test/OpenMP/for_reduction_codegen.cpp

// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --function-signature --check-globals --include-generated-funcs --replace-value-regex "__omp_offloading_[0-9a-z]+_[0-9a-z]+" "reduction_size[.].+[.]" "pl_cond[.].+[.|,]" --prefix-filecheck-ir-name _ --global-value-regex ".gomp_critical_user[a-zA-Z_.]+"
// RUN: %clang_cc1 -verify -Wno-vla -fopenmp -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix=CHECK1
// RUN: %clang_cc1 -fopenmp -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix=CHECK1
// RUN: %clang_cc1 -verify -Wno-vla -fopenmp -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix=CHECK3
// RUN: %clang_cc1 -verify -Wno-vla -fopenmp -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck -allow-deprecated-dag-overlap %s --check-prefix=CHECK4

// RUN: %clang_cc1 -verify -Wno-vla -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -fopenmp-simd -x c++ -std=c++11 -triple x86_64-apple-darwin10 -emit-pch -o %t %s
// RUN: %clang_cc1 -fopenmp-simd -x c++ -triple x86_64-apple-darwin10 -std=c++11 -include-pch %t -verify -Wno-vla %s -emit-llvm -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -verify -Wno-vla -fopenmp-simd -x c++ -std=c++11 -DLAMBDA -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// RUN: %clang_cc1 -verify -Wno-vla -fopenmp-simd -x c++ -fblocks -DBLOCKS -triple x86_64-apple-darwin10 -emit-llvm %s -o - | FileCheck %s --implicit-check-not="{{__kmpc|__tgt}}"
// expected-no-diagnostics
#ifndef HEADER
#define HEADER

volatile double g, g_orig;
volatile double &g1 = g_orig;

template <class T>
struct S {
  T f;
  S(T a) : f(a + g) {}
  S() : f(g) {}
  operator T() { return T(); }
  S &operator&(const S &) { return *this; }
  ~S() {}
};


template <typename T, int length>
T tmain() {
  T t;
  S<T> test;
  T t_var = T(), t_var1;
  T vec[] = {1, 2};
  S<T> s_arr[] = {1, 2};
  S<T> &var = test;
  S<T> var1;
  S<T> arr[length];
#pragma omp parallel
#pragma omp for reduction(+:t_var) reduction(&:var) reduction(&& : var1) reduction(min: t_var1) nowait
  for (int i = 0; i < 2; ++i) {
    vec[i] = t_var;
    s_arr[i] = var;
  }
#pragma omp parallel
#pragma omp for reduction(&& : t_var)
  for (int i = 0; i < 2; ++i) {
    vec[i] = t_var;
    s_arr[i] = var;
  }
#pragma omp parallel
#pragma omp for reduction(+ : arr[1:length-2])
  for (int i = 0; i < 2; ++i) {
    vec[i] = t_var;
    s_arr[i] = var;
  }
  return T();
}

extern S<float> **foo();

int main() {
#ifdef LAMBDA
  [&]() {
#pragma omp parallel
#pragma omp for reduction(+:g, g1)
    for (int i = 0; i < 2; ++i) {

    // Reduction list for runtime.

    g = 1;
    g1 = 1;

    [&]() {
      g = 2;
      g1 = 2;
    }();
  }
  }();
  return 0;
#elif defined(BLOCKS)
  ^{
#pragma omp parallel
#pragma omp for reduction(-:g, g1)
    for (int i = 0; i < 2; ++i)  {

    // Reduction list for runtime.

    g = 1;
    g1 = 1;

    ^{
      g = 2;
      g1 = 2;
    }();
  }
  }();
  return 0;
#else
  S<float> test;
  float t_var = 0, t_var1;
  int vec[] = {1, 2};
  S<float> s_arr[] = {1, 2, 3, 4};
  S<float> &var = test;
  S<float> var1, arrs[10][4];
  S<float> **var2 = foo();
  S<float> vvar2[5];
  S<float> (&var3)[4] = s_arr;
#pragma omp parallel
#pragma omp for reduction(+:t_var) reduction(&:var) reduction(&& : var1) reduction(min: t_var1)
  for (int i = 0; i < 2; ++i) {
    vec[i] = t_var;
    s_arr[i] = var;
  }
  int arr[10][vec[1]];
#pragma omp parallel for reduction(+:arr[1][:vec[1]]) reduction(&:arrs[1:vec[1]][1:2])
  for (int i = 0; i < 10; ++i)
    ++arr[1][i];
#pragma omp parallel
#pragma omp for reduction(+:arr) reduction(&:arrs)
  for (int i = 0; i < 10; ++i)
    ++arr[1][i];
  // arr is a VLA, but the array section has constant length so we can generate a constant sized array!
#pragma omp parallel
#pragma omp for reduction(+:arr[1][0:2])
  for (int i = 0; i < 10; ++i)
    ++arr[1][i];
#pragma omp parallel
#pragma omp for reduction(& : var2[0 : 5][1 : 6])
  for (int i = 0; i < 10; ++i)
    ;
#pragma omp parallel
#pragma omp for reduction(& : var2[1][1 : 6])
  for (int i = 0; i < 10; ++i)
    ;
#pragma omp parallel
#pragma omp for reduction(& : var2[1 : 1][1 : 6])
  for (int i = 0; i < 10; ++i)
    ;
#pragma omp parallel
#pragma omp for reduction(& : var2[1 : 1][1])
  for (int i = 0; i < 10; ++i)
    ;
#pragma omp parallel
#pragma omp for reduction(& : vvar2[0 : 5])
  for (int i = 0; i < 10; ++i)
    ;
#pragma omp parallel
#pragma omp for reduction(& : var3[1 : 2])
  for (int i = 0; i < 10; ++i)
    ;
#pragma omp parallel
#pragma omp for reduction(& : var3[ : 2])
  for (int i = 0; i < 10; ++i)
    ;
  // TODO: The compiler should also be able to generate a constant sized array in this case!
#pragma omp parallel
#pragma omp for reduction(& : var3[2 : ])
  for (int i = 0; i < 10; ++i)
    ;
#pragma omp parallel
#pragma omp for reduction(& : var3)
  for (int i = 0; i < 10; ++i)
    ;
  return tmain<int, 42>();
#endif
}


// Reduction list for runtime.


// For + reduction operation initial value of private variable is 0.


// For & reduction operation initial value of private variable is ones in all bits.

// For && reduction operation initial value of private variable is 1.0.

// For min reduction operation initial value of private variable is largest repesentable value.


// Skip checks for internal operations.

// ptr RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};


// res = __kmpc_reduce(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>);


// switch(res)

// case 1:
// t_var += t_var_reduction;

// var = var.operator &(var_reduction);

// var1 = var1.operator &&(var1_reduction);

// t_var1 = min(t_var1, t_var1_reduction);

// __kmpc_end_reduce(<loc>, <gtid>, &<lock>);

// break;

// case 2:
// t_var += t_var_reduction;

// var = var.operator &(var_reduction);

// var1 = var1.operator &&(var1_reduction);

// t_var1 = min(t_var1, t_var1_reduction);

// __kmpc_end_reduce(<loc>, <gtid>, &<lock>);

// break;


// void reduce_func(ptr lhs[<n>], ptr rhs[<n>]) {
//  *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]);
//  ...
//  *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1],
//  *(Type<n>-1*)rhs[<n>-1]);
// }
// t_var_lhs = (ptr)lhs[0];
// t_var_rhs = (ptr)rhs[0];

// var_lhs = (Sptr)lhs[1];
// var_rhs = (Sptr)rhs[1];

// var1_lhs = (Sptr)lhs[2];
// var1_rhs = (Sptr)rhs[2];

// t_var1_lhs = (ptr)lhs[3];
// t_var1_rhs = (ptr)rhs[3];

// t_var_lhs += t_var_rhs;

// var_lhs = var_lhs.operator &(var_rhs);

// var1_lhs = var1_lhs.operator &&(var1_rhs);

// t_var1_lhs = min(t_var1_lhs, t_var1_rhs);


// Reduction list for runtime.


// Check initialization of private copy.


// Check initialization of private copy.

// Skip checks for internal operations.

// ptr RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};


// res = __kmpc_reduce(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>);


// switch(res)

// case 1:

// arr[:] += arr_reduction[:];

// arrs[:] = var.operator &(arrs_reduction[:]);

// __kmpc_end_reduce(<loc>, <gtid>, &<lock>);

// break;

// case 2:

// arr[:] += arr_reduction[:];

// arrs[:] = var.operator &(arrs_reduction[:]);

// break;

// Check destruction of private copy.


// void reduce_func(ptr lhs[<n>], ptr rhs[<n>]) {
//  *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]);
//  ...
//  *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1],
//  *(Type<n>-1*)rhs[<n>-1]);
// }
// arr_rhs = (int*)rhs[0];
// arr_lhs = (int*)lhs[0];

// arr_size = (size_t)lhs[1];

// arrs_rhs = (Sptr)rhs[2];
// arrs_lhs = (Sptr)lhs[2];

// arrs_size = (size_t)lhs[3];

// arr_lhs[:] += arr_rhs[:];

// arrs_lhs = arrs_lhs.operator &(arrs_rhs);


// Reduction list for runtime.


// Check initialization of private copy.

// Check initialization of private copy.

// Skip checks for internal operations.

// ptr RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};


// res = __kmpc_reduce(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>);


// switch(res)

// case 1:

// arr[:] += arr_reduction[:];

// arrs[:] = var.operator &(arrs_reduction[:]);

// __kmpc_end_reduce(<loc>, <gtid>, &<lock>);

// break;

// case 2:

// arr[:] += arr_reduction[:];

// arrs[:] = var.operator &(arrs_reduction[:]);

// break;

// Check destruction of private copy.


// void reduce_func(ptr lhs[<n>], ptr rhs[<n>]) {
//  *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]);
//  ...
//  *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1],
//  *(Type<n>-1*)rhs[<n>-1]);
// }
// arr_rhs = (int*)rhs[0];
// arr_lhs = (int*)lhs[0];

// arr_size = (size_t)lhs[1];

// arrs_rhs = (Sptr)rhs[2];
// arrs_lhs = (Sptr)lhs[2];

// arr_lhs[:] += arr_rhs[:];

// arrs_lhs = arrs_lhs.operator &(arrs_rhs);


// Reduction list for runtime.


// Reduction list for runtime.


// Reduction list for runtime.


// Reduction list for runtime.


// Reduction list for runtime.


// Reduction list for runtime.


// Reduction list for runtime.


// Reduction list for runtime.


// Reduction list for runtime.


// Reduction list for runtime.


// Not interested in this one:

// Reduction list for runtime.


// For + reduction operation initial value of private variable is 0.

// For & reduction operation initial value of private variable is ones in all bits.

// For && reduction operation initial value of private variable is 1.0.

// For min reduction operation initial value of private variable is largest repesentable value.

// Skip checks for internal operations.

// ptr RedList[<n>] = {<ReductionVars>[0], ..., <ReductionVars>[<n>-1]};


// res = __kmpc_reduce_nowait(<loc>, <gtid>, <n>, sizeof(RedList), RedList, reduce_func, &<lock>);


// switch(res)

// case 1:
// t_var += t_var_reduction;

// var = var.operator &(var_reduction);

// var1 = var1.operator &&(var1_reduction);

// t_var1 = min(t_var1, t_var1_reduction);

// __kmpc_end_reduce_nowait(<loc>, <gtid>, &<lock>);

// break;

// case 2:
// t_var += t_var_reduction;

// var = var.operator &(var_reduction);

// var1 = var1.operator &&(var1_reduction);

// t_var1 = min(t_var1, t_var1_reduction);

// break;

// void reduce_func(ptr lhs[<n>], ptr rhs[<n>]) {
//  *(Type0*)lhs[0] = ReductionOperation0(*(Type0*)lhs[0], *(Type0*)rhs[0]);
//  ...
//  *(Type<n>-1*)lhs[<n>-1] = ReductionOperation<n>-1(*(Type<n>-1*)lhs[<n>-1],
//  *(Type<n>-1*)rhs[<n>-1]);
// }
// t_var_lhs = (iptr)lhs[0];
// t_var_rhs = (iptr)rhs[0];

// var_lhs = (Sptr)lhs[1];
// var_rhs = (Sptr)rhs[1];

// var1_lhs = (Sptr)lhs[2];
// var1_rhs = (Sptr)rhs[2];

// t_var1_lhs = (iptr)lhs[3];
// t_var1_rhs = (iptr)rhs[3];

// t_var_lhs += t_var_rhs;

// var_lhs = var_lhs.operator &(var_rhs);

// var1_lhs = var1_lhs.operator &&(var1_rhs);

// t_var1_lhs = min(t_var1_lhs, t_var1_rhs);


// Reduction list for runtime.


#endif
//.
// CHECK1: @.gomp_critical_user_.reduction.var = common global [8 x i32] zeroinitializer, align 8
// CHECK1: @.gomp_critical_user_.atomic_reduction.var = common global [8 x i32] zeroinitializer, align 8
//.
// CHECK3: @.gomp_critical_user_.reduction.var = common global [8 x i32] zeroinitializer, align 8
//.
// CHECK4: @.gomp_critical_user_.reduction.var = common global [8 x i32] zeroinitializer, align 8
//.
// CHECK1-LABEL: define {{[^@]+}}@main
// CHECK1-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    [[T_VAR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    [[T_VAR1:%.*]] = alloca float, align 4
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [4 x %struct.S], align 16
// CHECK1-NEXT:    [[VAR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR1:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK1-NEXT:    [[ARRS:%.*]] = alloca [10 x [4 x %struct.S]], align 16
// CHECK1-NEXT:    [[VAR2:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VVAR2:%.*]] = alloca [5 x %struct.S], align 16
// CHECK1-NEXT:    [[VAR3:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK1-NEXT:    store float 0.000000e+00, ptr [[T_VAR]], align 4
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const.main.vec, i64 8, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) [[S_ARR]], float noundef 1.000000e+00)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[S_ARR]], i64 1
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], float noundef 2.000000e+00)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT1:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[S_ARR]], i64 2
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT1]], float noundef 3.000000e+00)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT2:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[S_ARR]], i64 3
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT2]], float noundef 4.000000e+00)
// CHECK1-NEXT:    store ptr [[TEST]], ptr [[VAR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR1]])
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [4 x %struct.S]], ptr [[ARRS]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i64 40
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1:       arrayctor.loop:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1:       arrayctor.cont:
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef ptr @_Z3foov()
// CHECK1-NEXT:    store ptr [[CALL]], ptr [[VAR2]], align 8
// CHECK1-NEXT:    [[ARRAY_BEGIN3:%.*]] = getelementptr inbounds [5 x %struct.S], ptr [[VVAR2]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END4:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN3]], i64 5
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP5:%.*]]
// CHECK1:       arrayctor.loop5:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR6:%.*]] = phi ptr [ [[ARRAY_BEGIN3]], [[ARRAYCTOR_CONT]] ], [ [[ARRAYCTOR_NEXT7:%.*]], [[ARRAYCTOR_LOOP5]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR6]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT7]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYCTOR_CUR6]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE8:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT7]], [[ARRAYCTOR_END4]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE8]], label [[ARRAYCTOR_CONT9:%.*]], label [[ARRAYCTOR_LOOP5]]
// CHECK1:       arrayctor.cont9:
// CHECK1-NEXT:    store ptr [[S_ARR]], ptr [[VAR3]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VAR]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3:[0-9]+]], i32 6, ptr @main.omp_outlined, ptr [[T_VAR]], ptr [[TMP0]], ptr [[VAR1]], ptr [[T_VAR1]], ptr [[VEC]], ptr [[S_ARR]])
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[VEC]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP1:%.*]] = load i32, ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = zext i32 [[TMP1]] to i64
// CHECK1-NEXT:    [[TMP3:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK1-NEXT:    store ptr [[TMP3]], ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = mul nuw i64 10, [[TMP2]]
// CHECK1-NEXT:    [[VLA:%.*]] = alloca i32, i64 [[TMP4]], align 16
// CHECK1-NEXT:    store i64 [[TMP2]], ptr [[__VLA_EXPR0]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @main.omp_outlined.1, i64 10, i64 [[TMP2]], ptr [[VLA]], ptr [[VEC]], ptr [[ARRS]])
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @main.omp_outlined.2, i64 10, i64 [[TMP2]], ptr [[VLA]], ptr [[ARRS]])
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 3, ptr @main.omp_outlined.3, i64 10, i64 [[TMP2]], ptr [[VLA]])
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 1, ptr @main.omp_outlined.4, ptr [[VAR2]])
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 1, ptr @main.omp_outlined.5, ptr [[VAR2]])
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 1, ptr @main.omp_outlined.6, ptr [[VAR2]])
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 1, ptr @main.omp_outlined.7, ptr [[VAR2]])
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 1, ptr @main.omp_outlined.8, ptr [[VVAR2]])
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[VAR3]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 1, ptr @main.omp_outlined.9, ptr [[TMP5]])
// CHECK1-NEXT:    [[TMP6:%.*]] = load ptr, ptr [[VAR3]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 1, ptr @main.omp_outlined.10, ptr [[TMP6]])
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[VAR3]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 1, ptr @main.omp_outlined.11, ptr [[TMP7]])
// CHECK1-NEXT:    [[TMP8:%.*]] = load ptr, ptr [[VAR3]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 1, ptr @main.omp_outlined.12, ptr [[TMP8]])
// CHECK1-NEXT:    [[CALL10:%.*]] = call noundef i32 @_Z5tmainIiLi42EET_v()
// CHECK1-NEXT:    store i32 [[CALL10]], ptr [[RETVAL]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP9]])
// CHECK1-NEXT:    [[ARRAY_BEGIN11:%.*]] = getelementptr inbounds [5 x %struct.S], ptr [[VVAR2]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN11]], i64 5
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP10]], [[ARRAYCTOR_CONT9]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5:[0-9]+]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN11]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE12:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done12:
// CHECK1-NEXT:    [[ARRAY_BEGIN13:%.*]] = getelementptr inbounds [10 x [4 x %struct.S]], ptr [[ARRS]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN13]], i64 40
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY14:%.*]]
// CHECK1:       arraydestroy.body14:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST15:%.*]] = phi ptr [ [[TMP11]], [[ARRAYDESTROY_DONE12]] ], [ [[ARRAYDESTROY_ELEMENT16:%.*]], [[ARRAYDESTROY_BODY14]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT16]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST15]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT16]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE17:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT16]], [[ARRAY_BEGIN13]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE17]], label [[ARRAYDESTROY_DONE18:%.*]], label [[ARRAYDESTROY_BODY14]]
// CHECK1:       arraydestroy.done18:
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR1]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAY_BEGIN19:%.*]] = getelementptr inbounds [4 x %struct.S], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN19]], i64 4
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY20:%.*]]
// CHECK1:       arraydestroy.body20:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST21:%.*]] = phi ptr [ [[TMP12]], [[ARRAYDESTROY_DONE18]] ], [ [[ARRAYDESTROY_ELEMENT22:%.*]], [[ARRAYDESTROY_BODY20]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT22]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST21]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT22]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE23:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT22]], [[ARRAY_BEGIN19]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE23]], label [[ARRAYDESTROY_DONE24:%.*]], label [[ARRAYDESTROY_BODY20]]
// CHECK1:       arraydestroy.done24:
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR5]]
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP13]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1:[0-9]+]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIfEC2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC1Ef
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIfEC2Ef(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]], float noundef [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]], ptr noundef nonnull align 4 dereferenceable(8) [[VEC:%.*]], ptr noundef nonnull align 4 dereferenceable(16) [[S_ARR:%.*]]) #[[ATTR4:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR1_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[T_VAR1_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T_VAR3:%.*]] = alloca float, align 4
// CHECK1-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    [[_TMP5:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR16:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK1-NEXT:    [[T_VAR17:%.*]] = alloca float, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x ptr], align 8
// CHECK1-NEXT:    [[REF_TMP:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK1-NEXT:    [[REF_TMP23:%.*]] = alloca [[STRUCT_S]], align 4
// CHECK1-NEXT:    [[ATOMIC_TEMP:%.*]] = alloca float, align 4
// CHECK1-NEXT:    [[_TMP31:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[T_VAR]], ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR]], ptr [[VAR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR1]], ptr [[VAR1_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[T_VAR1]], ptr [[T_VAR1_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VEC]], ptr [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[S_ARR]], ptr [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[VAR1_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[T_VAR1_ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT:    store ptr [[TMP6]], ptr [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    store float 0.000000e+00, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR4]])
// CHECK1-NEXT:    store ptr [[VAR4]], ptr [[_TMP5]], align 8
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR16]])
// CHECK1-NEXT:    store float 0x47EFFFFFE0000000, ptr [[T_VAR17]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP10]], 1
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
// CHECK1-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP16:%.*]] = load float, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[CONV:%.*]] = fptosi float [[TMP16]] to i32
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[TMP4]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 [[CONV]], ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[TMP18:%.*]] = load ptr, ptr [[_TMP5]], align 8
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM9:%.*]] = sext i32 [[TMP19]] to i64
// CHECK1-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds [4 x %struct.S], ptr [[TMP5]], i64 0, i64 [[IDXPROM9]]
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[ARRAYIDX10]], ptr align 4 [[TMP18]], i64 4, i1 false)
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD11:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK1-NEXT:    store i32 [[ADD11]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP9]])
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[T_VAR3]], ptr [[TMP21]], align 8
// CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    store ptr [[VAR4]], ptr [[TMP22]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT:    store ptr [[VAR16]], ptr [[TMP23]], align 8
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3
// CHECK1-NEXT:    store ptr [[T_VAR17]], ptr [[TMP24]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP9]], i32 4, i64 32, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP25]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP26:%.*]] = load float, ptr [[TMP0]], align 4
// CHECK1-NEXT:    [[TMP27:%.*]] = load float, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[ADD12:%.*]] = fadd float [[TMP26]], [[TMP27]]
// CHECK1-NEXT:    store float [[ADD12]], ptr [[TMP0]], align 4
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[TMP7]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR4]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP7]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[CALL13:%.*]] = call noundef float @_ZN1SIfEcvfEv(ptr noundef nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK1-NEXT:    [[TOBOOL:%.*]] = fcmp une float [[CALL13]], 0.000000e+00
// CHECK1-NEXT:    br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]]
// CHECK1:       land.rhs:
// CHECK1-NEXT:    [[CALL14:%.*]] = call noundef float @_ZN1SIfEcvfEv(ptr noundef nonnull align 4 dereferenceable(4) [[VAR16]])
// CHECK1-NEXT:    [[TOBOOL15:%.*]] = fcmp une float [[CALL14]], 0.000000e+00
// CHECK1-NEXT:    br label [[LAND_END]]
// CHECK1:       land.end:
// CHECK1-NEXT:    [[TMP28:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE1]] ], [ [[TOBOOL15]], [[LAND_RHS]] ]
// CHECK1-NEXT:    [[CONV16:%.*]] = uitofp i1 [[TMP28]] to float
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]], float noundef [[CONV16]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP2]], ptr align 4 [[REF_TMP]], i64 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]]
// CHECK1-NEXT:    [[TMP29:%.*]] = load float, ptr [[TMP3]], align 4
// CHECK1-NEXT:    [[TMP30:%.*]] = load float, ptr [[T_VAR17]], align 4
// CHECK1-NEXT:    [[CMP17:%.*]] = fcmp olt float [[TMP29]], [[TMP30]]
// CHECK1-NEXT:    br i1 [[CMP17]], label [[COND_TRUE18:%.*]], label [[COND_FALSE19:%.*]]
// CHECK1:       cond.true18:
// CHECK1-NEXT:    [[TMP31:%.*]] = load float, ptr [[TMP3]], align 4
// CHECK1-NEXT:    br label [[COND_END20:%.*]]
// CHECK1:       cond.false19:
// CHECK1-NEXT:    [[TMP32:%.*]] = load float, ptr [[T_VAR17]], align 4
// CHECK1-NEXT:    br label [[COND_END20]]
// CHECK1:       cond.end20:
// CHECK1-NEXT:    [[COND21:%.*]] = phi float [ [[TMP31]], [[COND_TRUE18]] ], [ [[TMP32]], [[COND_FALSE19]] ]
// CHECK1-NEXT:    store float [[COND21]], ptr [[TMP3]], align 4
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP9]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP33:%.*]] = load float, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TMP34:%.*]] = atomicrmw fadd ptr [[TMP0]], float [[TMP33]] monotonic, align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP9]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL22:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[TMP7]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR4]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP7]], ptr align 4 [[CALL22]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP9]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP9]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL24:%.*]] = call noundef float @_ZN1SIfEcvfEv(ptr noundef nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK1-NEXT:    [[TOBOOL25:%.*]] = fcmp une float [[CALL24]], 0.000000e+00
// CHECK1-NEXT:    br i1 [[TOBOOL25]], label [[LAND_RHS26:%.*]], label [[LAND_END29:%.*]]
// CHECK1:       land.rhs26:
// CHECK1-NEXT:    [[CALL27:%.*]] = call noundef float @_ZN1SIfEcvfEv(ptr noundef nonnull align 4 dereferenceable(4) [[VAR16]])
// CHECK1-NEXT:    [[TOBOOL28:%.*]] = fcmp une float [[CALL27]], 0.000000e+00
// CHECK1-NEXT:    br label [[LAND_END29]]
// CHECK1:       land.end29:
// CHECK1-NEXT:    [[TMP35:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE2]] ], [ [[TOBOOL28]], [[LAND_RHS26]] ]
// CHECK1-NEXT:    [[CONV30:%.*]] = uitofp i1 [[TMP35]] to float
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP23]], float noundef [[CONV30]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP2]], ptr align 4 [[REF_TMP23]], i64 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP23]]) #[[ATTR5]]
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP9]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP36:%.*]] = load float, ptr [[T_VAR17]], align 4
// CHECK1-NEXT:    [[ATOMIC_LOAD:%.*]] = load atomic i32, ptr [[TMP3]] monotonic, align 4
// CHECK1-NEXT:    br label [[ATOMIC_CONT:%.*]]
// CHECK1:       atomic_cont:
// CHECK1-NEXT:    [[TMP37:%.*]] = phi i32 [ [[ATOMIC_LOAD]], [[LAND_END29]] ], [ [[TMP45:%.*]], [[COND_END35:%.*]] ]
// CHECK1-NEXT:    [[TMP38:%.*]] = bitcast i32 [[TMP37]] to float
// CHECK1-NEXT:    store float [[TMP38]], ptr [[_TMP31]], align 4
// CHECK1-NEXT:    [[TMP39:%.*]] = load float, ptr [[_TMP31]], align 4
// CHECK1-NEXT:    [[TMP40:%.*]] = load float, ptr [[T_VAR17]], align 4
// CHECK1-NEXT:    [[CMP32:%.*]] = fcmp olt float [[TMP39]], [[TMP40]]
// CHECK1-NEXT:    br i1 [[CMP32]], label [[COND_TRUE33:%.*]], label [[COND_FALSE34:%.*]]
// CHECK1:       cond.true33:
// CHECK1-NEXT:    [[TMP41:%.*]] = load float, ptr [[_TMP31]], align 4
// CHECK1-NEXT:    br label [[COND_END35]]
// CHECK1:       cond.false34:
// CHECK1-NEXT:    [[TMP42:%.*]] = load float, ptr [[T_VAR17]], align 4
// CHECK1-NEXT:    br label [[COND_END35]]
// CHECK1:       cond.end35:
// CHECK1-NEXT:    [[COND36:%.*]] = phi float [ [[TMP41]], [[COND_TRUE33]] ], [ [[TMP42]], [[COND_FALSE34]] ]
// CHECK1-NEXT:    store float [[COND36]], ptr [[ATOMIC_TEMP]], align 4
// CHECK1-NEXT:    [[TMP43:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4
// CHECK1-NEXT:    [[TMP44:%.*]] = cmpxchg ptr [[TMP3]], i32 [[TMP37]], i32 [[TMP43]] monotonic monotonic, align 4
// CHECK1-NEXT:    [[TMP45]] = extractvalue { i32, i1 } [[TMP44]], 0
// CHECK1-NEXT:    [[TMP46:%.*]] = extractvalue { i32, i1 } [[TMP44]], 1
// CHECK1-NEXT:    br i1 [[TMP46]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
// CHECK1:       atomic_exit:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP9]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR16]]) #[[ATTR5]]
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR5]]
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4:[0-9]+]], i32 [[TMP9]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6:[0-9]+]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[REF_TMP:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP15:%.*]] = load ptr, ptr [[TMP14]], align 8
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP3]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 8
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP19:%.*]] = load ptr, ptr [[TMP18]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = load float, ptr [[TMP7]], align 4
// CHECK1-NEXT:    [[TMP21:%.*]] = load float, ptr [[TMP5]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = fadd float [[TMP20]], [[TMP21]]
// CHECK1-NEXT:    store float [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[TMP11]], ptr noundef nonnull align 4 dereferenceable(4) [[TMP9]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP11]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[CALL2:%.*]] = call noundef float @_ZN1SIfEcvfEv(ptr noundef nonnull align 4 dereferenceable(4) [[TMP15]])
// CHECK1-NEXT:    [[TOBOOL:%.*]] = fcmp une float [[CALL2]], 0.000000e+00
// CHECK1-NEXT:    br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]]
// CHECK1:       land.rhs:
// CHECK1-NEXT:    [[CALL3:%.*]] = call noundef float @_ZN1SIfEcvfEv(ptr noundef nonnull align 4 dereferenceable(4) [[TMP13]])
// CHECK1-NEXT:    [[TOBOOL4:%.*]] = fcmp une float [[CALL3]], 0.000000e+00
// CHECK1-NEXT:    br label [[LAND_END]]
// CHECK1:       land.end:
// CHECK1-NEXT:    [[TMP22:%.*]] = phi i1 [ false, [[ENTRY:%.*]] ], [ [[TOBOOL4]], [[LAND_RHS]] ]
// CHECK1-NEXT:    [[CONV:%.*]] = uitofp i1 [[TMP22]] to float
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ef(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]], float noundef [[CONV]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP15]], ptr align 4 [[REF_TMP]], i64 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]]
// CHECK1-NEXT:    [[TMP23:%.*]] = load float, ptr [[TMP19]], align 4
// CHECK1-NEXT:    [[TMP24:%.*]] = load float, ptr [[TMP17]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = fcmp olt float [[TMP23]], [[TMP24]]
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    [[TMP25:%.*]] = load float, ptr [[TMP19]], align 4
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP26:%.*]] = load float, ptr [[TMP17]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi float [ [[TMP25]], [[COND_TRUE]] ], [ [[TMP26]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store float [[COND]], ptr [[TMP19]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEanERKS0_
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[TMP0:%.*]]) #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    ret ptr [[THIS1]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEcvfEv
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    ret float 0.000000e+00
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED1Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIfED2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR5]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.1
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARR:%.*]], ptr noundef nonnull align 4 dereferenceable(8) [[VEC:%.*]], ptr noundef nonnull align 4 dereferenceable(160) [[ARRS:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[ARRS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[__VLA_EXPR1:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK1-NEXT:    store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK1-NEXT:    store ptr [[ARR]], ptr [[ARR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VEC]], ptr [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[ARRS]], ptr [[ARRS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[ARRS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP5]]
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAYIDX]], i64 0
// CHECK1-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds [2 x i32], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[ARRAYIDX4]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = sext i32 [[TMP6]] to i64
// CHECK1-NEXT:    [[LB_ADD_LEN:%.*]] = add nsw i64 -1, [[TMP7]]
// CHECK1-NEXT:    [[TMP8:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP8]]
// CHECK1-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAYIDX5]], i64 [[LB_ADD_LEN]]
// CHECK1-NEXT:    [[TMP9:%.*]] = ptrtoint ptr [[ARRAYIDX6]] to i64
// CHECK1-NEXT:    [[TMP10:%.*]] = ptrtoint ptr [[ARRAYIDX3]] to i64
// CHECK1-NEXT:    [[TMP11:%.*]] = sub i64 [[TMP9]], [[TMP10]]
// CHECK1-NEXT:    [[TMP12:%.*]] = sdiv exact i64 [[TMP11]], ptrtoint (ptr getelementptr (i32, ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP13:%.*]] = add nuw i64 [[TMP12]], 1
// CHECK1-NEXT:    [[TMP14:%.*]] = mul nuw i64 [[TMP13]], ptrtoint (ptr getelementptr (i32, ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP15:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK1-NEXT:    store ptr [[TMP15]], ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT:    [[VLA7:%.*]] = alloca i32, i64 [[TMP13]], align 16
// CHECK1-NEXT:    store i64 [[TMP13]], ptr [[__VLA_EXPR0]], align 8
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr i32, ptr [[VLA7]], i64 [[TMP13]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[VLA7]], [[TMP16]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[VLA7]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP16]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP17:%.*]] = ptrtoint ptr [[TMP2]] to i64
// CHECK1-NEXT:    [[TMP18:%.*]] = ptrtoint ptr [[ARRAYIDX3]] to i64
// CHECK1-NEXT:    [[TMP19:%.*]] = sub i64 [[TMP17]], [[TMP18]]
// CHECK1-NEXT:    [[TMP20:%.*]] = sdiv exact i64 [[TMP19]], ptrtoint (ptr getelementptr (i32, ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr i32, ptr [[VLA7]], i64 [[TMP20]]
// CHECK1-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds nuw [10 x [4 x %struct.S]], ptr [[TMP4]], i64 0, i64 1
// CHECK1-NEXT:    [[ARRAYDECAY:%.*]] = getelementptr inbounds [4 x %struct.S], ptr [[ARRAYIDX8]], i64 0, i64 0
// CHECK1-NEXT:    [[ARRAYIDX9:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[ARRAYDECAY]], i64 1
// CHECK1-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds [2 x i32], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, ptr [[ARRAYIDX10]], align 4
// CHECK1-NEXT:    [[TMP23:%.*]] = sext i32 [[TMP22]] to i64
// CHECK1-NEXT:    [[LB_ADD_LEN11:%.*]] = add nsw i64 0, [[TMP23]]
// CHECK1-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds nuw [10 x [4 x %struct.S]], ptr [[TMP4]], i64 0, i64 [[LB_ADD_LEN11]]
// CHECK1-NEXT:    [[ARRAYDECAY13:%.*]] = getelementptr inbounds [4 x %struct.S], ptr [[ARRAYIDX12]], i64 0, i64 0
// CHECK1-NEXT:    [[ARRAYIDX14:%.*]] = getelementptr inbounds nuw [[STRUCT_S]], ptr [[ARRAYDECAY13]], i64 2
// CHECK1-NEXT:    [[TMP24:%.*]] = ptrtoint ptr [[ARRAYIDX14]] to i64
// CHECK1-NEXT:    [[TMP25:%.*]] = ptrtoint ptr [[ARRAYIDX9]] to i64
// CHECK1-NEXT:    [[TMP26:%.*]] = sub i64 [[TMP24]], [[TMP25]]
// CHECK1-NEXT:    [[TMP27:%.*]] = sdiv exact i64 [[TMP26]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP28:%.*]] = add nuw i64 [[TMP27]], 1
// CHECK1-NEXT:    [[TMP29:%.*]] = mul nuw i64 [[TMP28]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[VLA15:%.*]] = alloca [[STRUCT_S]], i64 [[TMP28]], align 16
// CHECK1-NEXT:    store i64 [[TMP28]], ptr [[__VLA_EXPR1]], align 8
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr [[STRUCT_S]], ptr [[VLA15]], i64 [[TMP28]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY16:%.*]] = icmp eq ptr [[VLA15]], [[TMP30]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY16]], label [[OMP_ARRAYINIT_DONE21:%.*]], label [[OMP_ARRAYINIT_BODY17:%.*]]
// CHECK1:       omp.arrayinit.body17:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST18:%.*]] = phi ptr [ [[VLA15]], [[OMP_ARRAYINIT_DONE]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT19:%.*]], [[OMP_ARRAYINIT_BODY17]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST18]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT19]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST18]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT19]], [[TMP30]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYINIT_DONE21]], label [[OMP_ARRAYINIT_BODY17]]
// CHECK1:       omp.arrayinit.done21:
// CHECK1-NEXT:    [[TMP31:%.*]] = ptrtoint ptr [[TMP4]] to i64
// CHECK1-NEXT:    [[TMP32:%.*]] = ptrtoint ptr [[ARRAYIDX9]] to i64
// CHECK1-NEXT:    [[TMP33:%.*]] = sub i64 [[TMP31]], [[TMP32]]
// CHECK1-NEXT:    [[TMP34:%.*]] = sdiv exact i64 [[TMP33]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP35:%.*]] = getelementptr [[STRUCT_S]], ptr [[VLA15]], i64 [[TMP34]]
// CHECK1-NEXT:    [[TMP36:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP37:%.*]] = load i32, ptr [[TMP36]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP37]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP38:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP38]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP39:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP39]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP40:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP40]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP41:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP42:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP22:%.*]] = icmp sle i32 [[TMP41]], [[TMP42]]
// CHECK1-NEXT:    br i1 [[CMP22]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP43:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP43]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP44:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX23:%.*]] = getelementptr inbounds i32, ptr [[TMP21]], i64 [[TMP44]]
// CHECK1-NEXT:    [[TMP45:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP45]] to i64
// CHECK1-NEXT:    [[ARRAYIDX24:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX23]], i64 [[IDXPROM]]
// CHECK1-NEXT:    [[TMP46:%.*]] = load i32, ptr [[ARRAYIDX24]], align 4
// CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP46]], 1
// CHECK1-NEXT:    store i32 [[INC]], ptr [[ARRAYIDX24]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP47:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD25:%.*]] = add nsw i32 [[TMP47]], 1
// CHECK1-NEXT:    store i32 [[ADD25]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP48:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP49:%.*]] = load i32, ptr [[TMP48]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP49]])
// CHECK1-NEXT:    [[TMP50:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VLA7]], ptr [[TMP50]], align 8
// CHECK1-NEXT:    [[TMP51:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP52:%.*]] = inttoptr i64 [[TMP13]] to ptr
// CHECK1-NEXT:    store ptr [[TMP52]], ptr [[TMP51]], align 8
// CHECK1-NEXT:    [[TMP53:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT:    store ptr [[VLA15]], ptr [[TMP53]], align 8
// CHECK1-NEXT:    [[TMP54:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP55:%.*]] = inttoptr i64 [[TMP28]] to ptr
// CHECK1-NEXT:    store ptr [[TMP55]], ptr [[TMP54]], align 8
// CHECK1-NEXT:    [[TMP56:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP57:%.*]] = load i32, ptr [[TMP56]], align 4
// CHECK1-NEXT:    [[TMP58:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB2]], i32 [[TMP57]], i32 2, i64 32, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.1.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP58]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP59:%.*]] = getelementptr i32, ptr [[ARRAYIDX3]], i64 [[TMP13]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX3]], [[TMP59]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE30:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA7]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST26:%.*]] = phi ptr [ [[ARRAYIDX3]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT28:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP60:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], align 4
// CHECK1-NEXT:    [[TMP61:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK1-NEXT:    [[ADD27:%.*]] = add nsw i32 [[TMP60]], [[TMP61]]
// CHECK1-NEXT:    store i32 [[ADD27]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT28]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST26]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE29:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT28]], [[TMP59]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE29]], label [[OMP_ARRAYCPY_DONE30]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done30:
// CHECK1-NEXT:    [[TMP62:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX9]], i64 [[TMP28]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY31:%.*]] = icmp eq ptr [[ARRAYIDX9]], [[TMP62]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY31]], label [[OMP_ARRAYCPY_DONE38:%.*]], label [[OMP_ARRAYCPY_BODY32:%.*]]
// CHECK1:       omp.arraycpy.body32:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST33:%.*]] = phi ptr [ [[VLA15]], [[OMP_ARRAYCPY_DONE30]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT36:%.*]], [[OMP_ARRAYCPY_BODY32]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST34:%.*]] = phi ptr [ [[ARRAYIDX9]], [[OMP_ARRAYCPY_DONE30]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT35:%.*]], [[OMP_ARRAYCPY_BODY32]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST34]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST33]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST34]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT35]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST34]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT36]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST33]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE37:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT35]], [[TMP62]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE37]], label [[OMP_ARRAYCPY_DONE38]], label [[OMP_ARRAYCPY_BODY32]]
// CHECK1:       omp.arraycpy.done38:
// CHECK1-NEXT:    call void @__kmpc_end_reduce_nowait(ptr @[[GLOB2]], i32 [[TMP57]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP63:%.*]] = getelementptr i32, ptr [[ARRAYIDX3]], i64 [[TMP13]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY39:%.*]] = icmp eq ptr [[ARRAYIDX3]], [[TMP63]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY39]], label [[OMP_ARRAYCPY_DONE46:%.*]], label [[OMP_ARRAYCPY_BODY40:%.*]]
// CHECK1:       omp.arraycpy.body40:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST41:%.*]] = phi ptr [ [[VLA7]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT44:%.*]], [[OMP_ARRAYCPY_BODY40]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST42:%.*]] = phi ptr [ [[ARRAYIDX3]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT43:%.*]], [[OMP_ARRAYCPY_BODY40]] ]
// CHECK1-NEXT:    [[TMP64:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST41]], align 4
// CHECK1-NEXT:    [[TMP65:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST42]], i32 [[TMP64]] monotonic, align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT43]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST42]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT44]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST41]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE45:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT43]], [[TMP63]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE45]], label [[OMP_ARRAYCPY_DONE46]], label [[OMP_ARRAYCPY_BODY40]]
// CHECK1:       omp.arraycpy.done46:
// CHECK1-NEXT:    [[TMP66:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX9]], i64 [[TMP28]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY47:%.*]] = icmp eq ptr [[ARRAYIDX9]], [[TMP66]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY47]], label [[OMP_ARRAYCPY_DONE55:%.*]], label [[OMP_ARRAYCPY_BODY48:%.*]]
// CHECK1:       omp.arraycpy.body48:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST49:%.*]] = phi ptr [ [[VLA15]], [[OMP_ARRAYCPY_DONE46]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT53:%.*]], [[OMP_ARRAYCPY_BODY48]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST50:%.*]] = phi ptr [ [[ARRAYIDX9]], [[OMP_ARRAYCPY_DONE46]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT52:%.*]], [[OMP_ARRAYCPY_BODY48]] ]
// CHECK1-NEXT:    [[TMP67:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP68:%.*]] = load i32, ptr [[TMP67]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP68]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL51:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST50]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST49]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST50]], ptr align 4 [[CALL51]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP68]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT52]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST50]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT53]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST49]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE54:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT52]], [[TMP66]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE54]], label [[OMP_ARRAYCPY_DONE55]], label [[OMP_ARRAYCPY_BODY48]]
// CHECK1:       omp.arraycpy.done55:
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[TMP69:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[VLA15]], i64 [[TMP28]]
// CHECK1-NEXT:    [[ARRAYDESTROY_ISEMPTY:%.*]] = icmp eq ptr [[VLA15]], [[TMP69]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_ISEMPTY]], label [[ARRAYDESTROY_DONE56:%.*]], label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP69]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[VLA15]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE56]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done56:
// CHECK1-NEXT:    [[TMP70:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP70]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.1.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = ptrtoint ptr [[TMP9]] to i64
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[TMP11]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP14:%.*]] = load ptr, ptr [[TMP13]], align 8
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP16:%.*]] = load ptr, ptr [[TMP15]], align 8
// CHECK1-NEXT:    [[TMP17:%.*]] = ptrtoint ptr [[TMP16]] to i64
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr i32, ptr [[TMP7]], i64 [[TMP10]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP18]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP19]], [[TMP20]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP18]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[TMP14]], i64 [[TMP17]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY3:%.*]] = icmp eq ptr [[TMP14]], [[TMP21]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY3]], label [[OMP_ARRAYCPY_DONE10:%.*]], label [[OMP_ARRAYCPY_BODY4:%.*]]
// CHECK1:       omp.arraycpy.body4:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST5:%.*]] = phi ptr [ [[TMP12]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT8:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST6:%.*]] = phi ptr [ [[TMP14]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT7:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST6]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST5]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST6]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT7]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST6]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT8]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST5]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE9:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT7]], [[TMP21]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_BODY4]]
// CHECK1:       omp.arraycpy.done10:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.2
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARR:%.*]], ptr noundef nonnull align 4 dereferenceable(160) [[ARRS:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[ARRS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[ARRS4:%.*]] = alloca [10 x [4 x %struct.S]], align 16
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [3 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK1-NEXT:    store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK1-NEXT:    store ptr [[ARR]], ptr [[ARR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[ARRS]], ptr [[ARRS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[ARRS_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP4:%.*]] = mul nuw i64 [[TMP0]], [[TMP1]]
// CHECK1-NEXT:    [[TMP5:%.*]] = mul nuw i64 [[TMP4]], 4
// CHECK1-NEXT:    [[TMP6:%.*]] = udiv exact i64 [[TMP5]], ptrtoint (ptr getelementptr (i32, ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP7:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK1-NEXT:    store ptr [[TMP7]], ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT:    [[VLA3:%.*]] = alloca i32, i64 [[TMP6]], align 16
// CHECK1-NEXT:    store i64 [[TMP6]], ptr [[__VLA_EXPR0]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr i32, ptr [[VLA3]], i64 [[TMP6]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[VLA3]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[VLA3]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [10 x [4 x %struct.S]], ptr [[ARRS4]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP9:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[ARRAY_BEGIN]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY5:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP9]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY5]], label [[OMP_ARRAYINIT_DONE10:%.*]], label [[OMP_ARRAYINIT_BODY6:%.*]]
// CHECK1:       omp.arrayinit.body6:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST7:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[OMP_ARRAYINIT_DONE]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYINIT_BODY6]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST7]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST7]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE9:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP9]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYINIT_DONE10]], label [[OMP_ARRAYINIT_BODY6]]
// CHECK1:       omp.arrayinit.done10:
// CHECK1-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[TMP10]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP11]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP12]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP13]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP14]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP11:%.*]] = icmp sle i32 [[TMP15]], [[TMP16]]
// CHECK1-NEXT:    br i1 [[CMP11]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP17]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP18:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[VLA3]], i64 [[TMP18]]
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP19]] to i64
// CHECK1-NEXT:    [[ARRAYIDX12:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX]], i64 [[IDXPROM]]
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[ARRAYIDX12]], align 4
// CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK1-NEXT:    store i32 [[INC]], ptr [[ARRAYIDX12]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD13:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK1-NEXT:    store i32 [[ADD13]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VLA3]], ptr [[TMP24]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP26:%.*]] = inttoptr i64 [[TMP6]] to ptr
// CHECK1-NEXT:    store ptr [[TMP26]], ptr [[TMP25]], align 8
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [3 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT:    store ptr [[ARRS4]], ptr [[TMP27]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP29:%.*]] = load i32, ptr [[TMP28]], align 4
// CHECK1-NEXT:    [[TMP30:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP29]], i32 2, i64 24, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.2.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP30]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP31:%.*]] = getelementptr i32, ptr [[TMP2]], i64 [[TMP6]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP2]], [[TMP31]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE18:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA3]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST14:%.*]] = phi ptr [ [[TMP2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT16:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP32:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST14]], align 4
// CHECK1-NEXT:    [[TMP33:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK1-NEXT:    [[ADD15:%.*]] = add nsw i32 [[TMP32]], [[TMP33]]
// CHECK1-NEXT:    store i32 [[ADD15]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST14]], align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT16]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST14]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE17:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT16]], [[TMP31]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done18:
// CHECK1-NEXT:    [[TMP34:%.*]] = getelementptr [[STRUCT_S]], ptr [[TMP3]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY19:%.*]] = icmp eq ptr [[TMP3]], [[TMP34]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY19]], label [[OMP_ARRAYCPY_DONE26:%.*]], label [[OMP_ARRAYCPY_BODY20:%.*]]
// CHECK1:       omp.arraycpy.body20:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST21:%.*]] = phi ptr [ [[ARRS4]], [[OMP_ARRAYCPY_DONE18]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT24:%.*]], [[OMP_ARRAYCPY_BODY20]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST22:%.*]] = phi ptr [ [[TMP3]], [[OMP_ARRAYCPY_DONE18]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT23:%.*]], [[OMP_ARRAYCPY_BODY20]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST22]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST21]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST22]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT23]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST22]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT24]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST21]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE25:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT23]], [[TMP34]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE25]], label [[OMP_ARRAYCPY_DONE26]], label [[OMP_ARRAYCPY_BODY20]]
// CHECK1:       omp.arraycpy.done26:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP29]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP35:%.*]] = getelementptr i32, ptr [[TMP2]], i64 [[TMP6]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY27:%.*]] = icmp eq ptr [[TMP2]], [[TMP35]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY27]], label [[OMP_ARRAYCPY_DONE34:%.*]], label [[OMP_ARRAYCPY_BODY28:%.*]]
// CHECK1:       omp.arraycpy.body28:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST29:%.*]] = phi ptr [ [[VLA3]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT32:%.*]], [[OMP_ARRAYCPY_BODY28]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST30:%.*]] = phi ptr [ [[TMP2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT31:%.*]], [[OMP_ARRAYCPY_BODY28]] ]
// CHECK1-NEXT:    [[TMP36:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST29]], align 4
// CHECK1-NEXT:    [[TMP37:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST30]], i32 [[TMP36]] monotonic, align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT31]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST30]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT32]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST29]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE33:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT31]], [[TMP35]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE33]], label [[OMP_ARRAYCPY_DONE34]], label [[OMP_ARRAYCPY_BODY28]]
// CHECK1:       omp.arraycpy.done34:
// CHECK1-NEXT:    [[TMP38:%.*]] = getelementptr [[STRUCT_S]], ptr [[TMP3]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY35:%.*]] = icmp eq ptr [[TMP3]], [[TMP38]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY35]], label [[OMP_ARRAYCPY_DONE43:%.*]], label [[OMP_ARRAYCPY_BODY36:%.*]]
// CHECK1:       omp.arraycpy.body36:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST37:%.*]] = phi ptr [ [[ARRS4]], [[OMP_ARRAYCPY_DONE34]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT41:%.*]], [[OMP_ARRAYCPY_BODY36]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST38:%.*]] = phi ptr [ [[TMP3]], [[OMP_ARRAYCPY_DONE34]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT40:%.*]], [[OMP_ARRAYCPY_BODY36]] ]
// CHECK1-NEXT:    [[TMP39:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP40:%.*]] = load i32, ptr [[TMP39]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP40]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL39:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST38]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST37]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST38]], ptr align 4 [[CALL39]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP40]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT40]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST38]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT41]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST37]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE42:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT40]], [[TMP38]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE42]], label [[OMP_ARRAYCPY_DONE43]], label [[OMP_ARRAYCPY_BODY36]]
// CHECK1:       omp.arraycpy.done43:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP29]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN44:%.*]] = getelementptr inbounds [10 x [4 x %struct.S]], ptr [[ARRS4]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN44]], i64 40
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP41]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN44]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE45:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done45:
// CHECK1-NEXT:    [[TMP42:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP42]])
// CHECK1-NEXT:    [[TMP43:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP44:%.*]] = load i32, ptr [[TMP43]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP44]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.2.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = ptrtoint ptr [[TMP9]] to i64
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[TMP11]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = getelementptr inbounds [3 x ptr], ptr [[TMP2]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP14:%.*]] = load ptr, ptr [[TMP13]], align 8
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr i32, ptr [[TMP7]], i64 [[TMP10]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP15]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP15]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[TMP14]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY3:%.*]] = icmp eq ptr [[TMP14]], [[TMP18]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY3]], label [[OMP_ARRAYCPY_DONE10:%.*]], label [[OMP_ARRAYCPY_BODY4:%.*]]
// CHECK1:       omp.arraycpy.body4:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST5:%.*]] = phi ptr [ [[TMP12]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT8:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST6:%.*]] = phi ptr [ [[TMP14]], [[OMP_ARRAYCPY_DONE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT7:%.*]], [[OMP_ARRAYCPY_BODY4]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST6]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST5]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST6]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT7]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST6]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT8]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST5]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE9:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT7]], [[TMP18]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_BODY4]]
// CHECK1:       omp.arraycpy.done10:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.3
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], i64 noundef [[VLA:%.*]], i64 noundef [[VLA1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[ARR:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VLA_ADDR:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[VLA_ADDR2:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[ARR6:%.*]] = alloca [1 x [2 x i32]], align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store i64 [[VLA]], ptr [[VLA_ADDR]], align 8
// CHECK1-NEXT:    store i64 [[VLA1]], ptr [[VLA_ADDR2]], align 8
// CHECK1-NEXT:    store ptr [[ARR]], ptr [[ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i64, ptr [[VLA_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load i64, ptr [[VLA_ADDR2]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[ARR_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP3:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP3]]
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAYIDX]], i64 0
// CHECK1-NEXT:    [[TMP4:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX4:%.*]] = getelementptr inbounds i32, ptr [[TMP2]], i64 [[TMP4]]
// CHECK1-NEXT:    [[ARRAYIDX5:%.*]] = getelementptr inbounds nuw i32, ptr [[ARRAYIDX4]], i64 1
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [1 x [2 x i32]], ptr [[ARR6]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr i32, ptr [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP5]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    store i32 0, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP6:%.*]] = ptrtoint ptr [[TMP2]] to i64
// CHECK1-NEXT:    [[TMP7:%.*]] = ptrtoint ptr [[ARRAYIDX3]] to i64
// CHECK1-NEXT:    [[TMP8:%.*]] = sub i64 [[TMP6]], [[TMP7]]
// CHECK1-NEXT:    [[TMP9:%.*]] = sdiv exact i64 [[TMP8]], ptrtoint (ptr getelementptr (i32, ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr i32, ptr [[ARR6]], i64 [[TMP9]]
// CHECK1-NEXT:    [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP12]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP13]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT:    br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP19:%.*]] = mul nsw i64 1, [[TMP1]]
// CHECK1-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds i32, ptr [[TMP10]], i64 [[TMP19]]
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP20]] to i64
// CHECK1-NEXT:    [[ARRAYIDX9:%.*]] = getelementptr inbounds i32, ptr [[ARRAYIDX8]], i64 [[IDXPROM]]
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[ARRAYIDX9]], align 4
// CHECK1-NEXT:    [[INC:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK1-NEXT:    store i32 [[INC]], ptr [[ARRAYIDX9]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD10:%.*]] = add nsw i32 [[TMP22]], 1
// CHECK1-NEXT:    store i32 [[ADD10]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP24]])
// CHECK1-NEXT:    [[TMP25:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[ARR6]], ptr [[TMP25]], align 8
// CHECK1-NEXT:    [[TMP26:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP27:%.*]] = load i32, ptr [[TMP26]], align 4
// CHECK1-NEXT:    [[TMP28:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP27]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.3.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP28]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP29:%.*]] = getelementptr i32, ptr [[ARRAYIDX3]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX3]], [[TMP29]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE15:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[ARR6]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST11:%.*]] = phi ptr [ [[ARRAYIDX3]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT13:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP30:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST11]], align 4
// CHECK1-NEXT:    [[TMP31:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK1-NEXT:    [[ADD12:%.*]] = add nsw i32 [[TMP30]], [[TMP31]]
// CHECK1-NEXT:    store i32 [[ADD12]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST11]], align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT13]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST11]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE14:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT13]], [[TMP29]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_DONE15]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done15:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP27]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP32:%.*]] = getelementptr i32, ptr [[ARRAYIDX3]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY16:%.*]] = icmp eq ptr [[ARRAYIDX3]], [[TMP32]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY16]], label [[OMP_ARRAYCPY_DONE23:%.*]], label [[OMP_ARRAYCPY_BODY17:%.*]]
// CHECK1:       omp.arraycpy.body17:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST18:%.*]] = phi ptr [ [[ARR6]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT21:%.*]], [[OMP_ARRAYCPY_BODY17]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST19:%.*]] = phi ptr [ [[ARRAYIDX3]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT20:%.*]], [[OMP_ARRAYCPY_BODY17]] ]
// CHECK1-NEXT:    [[TMP33:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST18]], align 4
// CHECK1-NEXT:    [[TMP34:%.*]] = atomicrmw add ptr [[OMP_ARRAYCPY_DESTELEMENTPAST19]], i32 [[TMP33]] monotonic, align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT20]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST19]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT21]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST18]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE22:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT20]], [[TMP32]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE22]], label [[OMP_ARRAYCPY_DONE23]], label [[OMP_ARRAYCPY_BODY17]]
// CHECK1:       omp.arraycpy.done23:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP27]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[TMP35:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP36:%.*]] = load i32, ptr [[TMP35]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP36]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.3.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr i32, ptr [[TMP7]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP9]], [[TMP10]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], align 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr i32, ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.4
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[VAR2:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR2_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[_TMP4:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP5:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR2]], ptr [[VAR2_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VAR2_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw ptr, ptr [[TMP1]], i64 0
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[ARRAYIDX]], align 8
// CHECK1-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[TMP2]], i64 1
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw ptr, ptr [[TMP3]], i64 4
// CHECK1-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[ARRAYIDX2]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw [[STRUCT_S]], ptr [[TMP4]], i64 6
// CHECK1-NEXT:    [[TMP5:%.*]] = ptrtoint ptr [[ARRAYIDX3]] to i64
// CHECK1-NEXT:    [[TMP6:%.*]] = ptrtoint ptr [[ARRAYIDX1]] to i64
// CHECK1-NEXT:    [[TMP7:%.*]] = sub i64 [[TMP5]], [[TMP6]]
// CHECK1-NEXT:    [[TMP8:%.*]] = sdiv exact i64 [[TMP7]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP9:%.*]] = add nuw i64 [[TMP8]], 1
// CHECK1-NEXT:    [[TMP10:%.*]] = mul nuw i64 [[TMP9]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP11:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK1-NEXT:    store ptr [[TMP11]], ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT:    [[VLA:%.*]] = alloca [[STRUCT_S]], i64 [[TMP9]], align 16
// CHECK1-NEXT:    store i64 [[TMP9]], ptr [[__VLA_EXPR0]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S]], ptr [[VLA]], i64 [[TMP9]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[VLA]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP12]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP13:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = load ptr, ptr [[TMP13]], align 8
// CHECK1-NEXT:    [[TMP15:%.*]] = ptrtoint ptr [[TMP14]] to i64
// CHECK1-NEXT:    [[TMP16:%.*]] = ptrtoint ptr [[ARRAYIDX1]] to i64
// CHECK1-NEXT:    [[TMP17:%.*]] = sub i64 [[TMP15]], [[TMP16]]
// CHECK1-NEXT:    [[TMP18:%.*]] = sdiv exact i64 [[TMP17]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP19:%.*]] = getelementptr [[STRUCT_S]], ptr [[VLA]], i64 [[TMP18]]
// CHECK1-NEXT:    store ptr [[_TMP5]], ptr [[_TMP4]], align 8
// CHECK1-NEXT:    store ptr [[TMP19]], ptr [[_TMP5]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP21]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP22]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP23]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP24]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP25]], [[TMP26]]
// CHECK1-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP27:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP27]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP28:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP28]], 1
// CHECK1-NEXT:    store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP29:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP30:%.*]] = load i32, ptr [[TMP29]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP30]])
// CHECK1-NEXT:    [[TMP31:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VLA]], ptr [[TMP31]], align 8
// CHECK1-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP33:%.*]] = inttoptr i64 [[TMP9]] to ptr
// CHECK1-NEXT:    store ptr [[TMP33]], ptr [[TMP32]], align 8
// CHECK1-NEXT:    [[TMP34:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP35:%.*]] = load i32, ptr [[TMP34]], align 4
// CHECK1-NEXT:    [[TMP36:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP35]], i32 1, i64 16, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.4.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP36]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP37:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX1]], i64 [[TMP9]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX1]], [[TMP37]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE11:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST8:%.*]] = phi ptr [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT9:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST8]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST8]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT9]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST8]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE10:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT9]], [[TMP37]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_DONE11]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done11:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP35]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP38:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX1]], i64 [[TMP9]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY12:%.*]] = icmp eq ptr [[ARRAYIDX1]], [[TMP38]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY12]], label [[OMP_ARRAYCPY_DONE20:%.*]], label [[OMP_ARRAYCPY_BODY13:%.*]]
// CHECK1:       omp.arraycpy.body13:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST14:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT18:%.*]], [[OMP_ARRAYCPY_BODY13]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST15:%.*]] = phi ptr [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT17:%.*]], [[OMP_ARRAYCPY_BODY13]] ]
// CHECK1-NEXT:    [[TMP39:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP40:%.*]] = load i32, ptr [[TMP39]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP40]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL16:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST15]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST14]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST15]], ptr align 4 [[CALL16]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP40]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT17]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT18]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST14]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE19:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT17]], [[TMP38]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE19]], label [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_BODY13]]
// CHECK1:       omp.arraycpy.done20:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP35]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[TMP41:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[VLA]], i64 [[TMP9]]
// CHECK1-NEXT:    [[ARRAYDESTROY_ISEMPTY:%.*]] = icmp eq ptr [[VLA]], [[TMP41]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_ISEMPTY]], label [[ARRAYDESTROY_DONE21:%.*]], label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP41]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[VLA]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE21]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done21:
// CHECK1-NEXT:    [[TMP42:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP42]])
// CHECK1-NEXT:    [[TMP43:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP44:%.*]] = load i32, ptr [[TMP43]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP44]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.4.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = ptrtoint ptr [[TMP9]] to i64
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[TMP7]], i64 [[TMP10]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP11]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP11]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.5
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[VAR2:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR2_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VAR24:%.*]] = alloca [1 x [6 x %struct.S]], align 16
// CHECK1-NEXT:    [[_TMP5:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP6:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR2]], ptr [[VAR2_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VAR2_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds ptr, ptr [[TMP1]], i64 1
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[ARRAYIDX]], align 8
// CHECK1-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[TMP2]], i64 1
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds ptr, ptr [[TMP3]], i64 1
// CHECK1-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[ARRAYIDX2]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw [[STRUCT_S]], ptr [[TMP4]], i64 6
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [1 x [6 x %struct.S]], ptr [[VAR24]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i64 6
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP5]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP6:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = ptrtoint ptr [[TMP7]] to i64
// CHECK1-NEXT:    [[TMP9:%.*]] = ptrtoint ptr [[ARRAYIDX1]] to i64
// CHECK1-NEXT:    [[TMP10:%.*]] = sub i64 [[TMP8]], [[TMP9]]
// CHECK1-NEXT:    [[TMP11:%.*]] = sdiv exact i64 [[TMP10]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S]], ptr [[VAR24]], i64 [[TMP11]]
// CHECK1-NEXT:    store ptr [[_TMP6]], ptr [[_TMP5]], align 8
// CHECK1-NEXT:    store ptr [[TMP12]], ptr [[_TMP6]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP14]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP15]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
// CHECK1-NEXT:    br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK1-NEXT:    store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VAR24]], ptr [[TMP24]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
// CHECK1-NEXT:    [[TMP27:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP26]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.5.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP27]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP28:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX1]], i64 6
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX1]], [[TMP28]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE12:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VAR24]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST9:%.*]] = phi ptr [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT10:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST9]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST9]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT10]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST9]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE11:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT10]], [[TMP28]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE11]], label [[OMP_ARRAYCPY_DONE12]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done12:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP26]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP29:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX1]], i64 6
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY13:%.*]] = icmp eq ptr [[ARRAYIDX1]], [[TMP29]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY13]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY14:%.*]]
// CHECK1:       omp.arraycpy.body14:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST15:%.*]] = phi ptr [ [[VAR24]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT19:%.*]], [[OMP_ARRAYCPY_BODY14]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi ptr [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT18:%.*]], [[OMP_ARRAYCPY_BODY14]] ]
// CHECK1-NEXT:    [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP31]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL17:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST16]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST15]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST16]], ptr align 4 [[CALL17]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP31]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT18]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT19]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST15]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT18]], [[TMP29]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY14]]
// CHECK1:       omp.arraycpy.done21:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP26]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN22:%.*]] = getelementptr inbounds [1 x [6 x %struct.S]], ptr [[VAR24]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN22]], i64 6
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP32]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN22]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE23:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done23:
// CHECK1-NEXT:    [[TMP33:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP34:%.*]] = load i32, ptr [[TMP33]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP34]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.5.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[TMP7]], i64 6
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.6
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[VAR2:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR2_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VAR24:%.*]] = alloca [1 x [6 x %struct.S]], align 16
// CHECK1-NEXT:    [[_TMP5:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP6:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR2]], ptr [[VAR2_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VAR2_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw ptr, ptr [[TMP1]], i64 1
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[ARRAYIDX]], align 8
// CHECK1-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[TMP2]], i64 1
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw ptr, ptr [[TMP3]], i64 1
// CHECK1-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[ARRAYIDX2]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw [[STRUCT_S]], ptr [[TMP4]], i64 6
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [1 x [6 x %struct.S]], ptr [[VAR24]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP5:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAY_BEGIN]], i64 6
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP5]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP5]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP6:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = ptrtoint ptr [[TMP7]] to i64
// CHECK1-NEXT:    [[TMP9:%.*]] = ptrtoint ptr [[ARRAYIDX1]] to i64
// CHECK1-NEXT:    [[TMP10:%.*]] = sub i64 [[TMP8]], [[TMP9]]
// CHECK1-NEXT:    [[TMP11:%.*]] = sdiv exact i64 [[TMP10]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr [[STRUCT_S]], ptr [[VAR24]], i64 [[TMP11]]
// CHECK1-NEXT:    store ptr [[_TMP6]], ptr [[_TMP5]], align 8
// CHECK1-NEXT:    store ptr [[TMP12]], ptr [[_TMP6]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP14]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP15]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP16]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP17]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP18]], [[TMP19]]
// CHECK1-NEXT:    br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP20]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP21]], 1
// CHECK1-NEXT:    store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP22:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, ptr [[TMP22]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP23]])
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VAR24]], ptr [[TMP24]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
// CHECK1-NEXT:    [[TMP27:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP26]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.6.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP27]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP28:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX1]], i64 6
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX1]], [[TMP28]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE12:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VAR24]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST9:%.*]] = phi ptr [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT10:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST9]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST9]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT10]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST9]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE11:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT10]], [[TMP28]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE11]], label [[OMP_ARRAYCPY_DONE12]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done12:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP26]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP29:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX1]], i64 6
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY13:%.*]] = icmp eq ptr [[ARRAYIDX1]], [[TMP29]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY13]], label [[OMP_ARRAYCPY_DONE21:%.*]], label [[OMP_ARRAYCPY_BODY14:%.*]]
// CHECK1:       omp.arraycpy.body14:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST15:%.*]] = phi ptr [ [[VAR24]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT19:%.*]], [[OMP_ARRAYCPY_BODY14]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST16:%.*]] = phi ptr [ [[ARRAYIDX1]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT18:%.*]], [[OMP_ARRAYCPY_BODY14]] ]
// CHECK1-NEXT:    [[TMP30:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP31:%.*]] = load i32, ptr [[TMP30]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP31]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL17:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST16]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST15]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST16]], ptr align 4 [[CALL17]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP31]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT18]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST16]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT19]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST15]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE20:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT18]], [[TMP29]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_DONE21]], label [[OMP_ARRAYCPY_BODY14]]
// CHECK1:       omp.arraycpy.done21:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP26]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN22:%.*]] = getelementptr inbounds [1 x [6 x %struct.S]], ptr [[VAR24]], i32 0, i32 0, i32 0
// CHECK1-NEXT:    [[TMP32:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN22]], i64 6
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP32]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN22]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE23:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done23:
// CHECK1-NEXT:    [[TMP33:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP34:%.*]] = load i32, ptr [[TMP33]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP34]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.6.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[TMP7]], i64 6
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.7
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 8 dereferenceable(8) [[VAR2:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR2_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VAR24:%.*]] = alloca [[STRUCT_S:%.*]], align 4
// CHECK1-NEXT:    [[_TMP5:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP6:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR2]], ptr [[VAR2_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VAR2_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw ptr, ptr [[TMP1]], i64 1
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[ARRAYIDX]], align 8
// CHECK1-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw [[STRUCT_S]], ptr [[TMP2]], i64 1
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[ARRAYIDX2:%.*]] = getelementptr inbounds nuw ptr, ptr [[TMP3]], i64 1
// CHECK1-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[ARRAYIDX2]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw [[STRUCT_S]], ptr [[TMP4]], i64 1
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR24]])
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP0]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = load ptr, ptr [[TMP5]], align 8
// CHECK1-NEXT:    [[TMP7:%.*]] = ptrtoint ptr [[TMP6]] to i64
// CHECK1-NEXT:    [[TMP8:%.*]] = ptrtoint ptr [[ARRAYIDX1]] to i64
// CHECK1-NEXT:    [[TMP9:%.*]] = sub i64 [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    [[TMP10:%.*]] = sdiv exact i64 [[TMP9]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr [[STRUCT_S]], ptr [[VAR24]], i64 [[TMP10]]
// CHECK1-NEXT:    store ptr [[_TMP6]], ptr [[_TMP5]], align 8
// CHECK1-NEXT:    store ptr [[TMP11]], ptr [[_TMP6]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP13]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP14]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP16]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP7:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
// CHECK1-NEXT:    br i1 [[CMP7]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD8:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK1-NEXT:    store i32 [[ADD8]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP13]])
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VAR24]], ptr [[TMP21]], align 8
// CHECK1-NEXT:    [[TMP22:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP13]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.7.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP22]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYIDX1]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR24]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[ARRAYIDX1]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP13]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP13]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL9:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYIDX1]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR24]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[ARRAYIDX1]], ptr align 4 [[CALL9]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP13]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP13]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR24]]) #[[ATTR5]]
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP13]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.7.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[TMP7]], ptr noundef nonnull align 4 dereferenceable(4) [[TMP5]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP7]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.8
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(20) [[VVAR2:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VVAR2_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VVAR22:%.*]] = alloca [5 x %struct.S], align 16
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VVAR2]], ptr [[VVAR2_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VVAR2_ADDR]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [5 x %struct.S], ptr [[TMP0]], i64 0, i64 0
// CHECK1-NEXT:    [[ARRAYIDX1:%.*]] = getelementptr inbounds nuw [5 x %struct.S], ptr [[TMP0]], i64 0, i64 4
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [5 x %struct.S], ptr [[VVAR22]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP1:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[ARRAY_BEGIN]], i64 5
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP1]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP1]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP2:%.*]] = ptrtoint ptr [[TMP0]] to i64
// CHECK1-NEXT:    [[TMP3:%.*]] = ptrtoint ptr [[ARRAYIDX]] to i64
// CHECK1-NEXT:    [[TMP4:%.*]] = sub i64 [[TMP2]], [[TMP3]]
// CHECK1-NEXT:    [[TMP5:%.*]] = sdiv exact i64 [[TMP4]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S]], ptr [[VVAR22]], i64 [[TMP5]]
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP8]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP9]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP10]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP11]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP3:%.*]] = icmp sle i32 [[TMP12]], [[TMP13]]
// CHECK1-NEXT:    br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP14]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD4:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK1-NEXT:    store i32 [[ADD4]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP17]])
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VVAR22]], ptr [[TMP18]], align 8
// CHECK1-NEXT:    [[TMP19:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[TMP19]], align 4
// CHECK1-NEXT:    [[TMP21:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP20]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.8.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP21]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX]], i64 5
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX]], [[TMP22]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE8:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VVAR22]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST5:%.*]] = phi ptr [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT6:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST5]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST5]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT6]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST5]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE7:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT6]], [[TMP22]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE7]], label [[OMP_ARRAYCPY_DONE8]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done8:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP20]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP23:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX]], i64 5
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY9:%.*]] = icmp eq ptr [[ARRAYIDX]], [[TMP23]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY9]], label [[OMP_ARRAYCPY_DONE17:%.*]], label [[OMP_ARRAYCPY_BODY10:%.*]]
// CHECK1:       omp.arraycpy.body10:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST11:%.*]] = phi ptr [ [[VVAR22]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT15:%.*]], [[OMP_ARRAYCPY_BODY10]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST12:%.*]] = phi ptr [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT14:%.*]], [[OMP_ARRAYCPY_BODY10]] ]
// CHECK1-NEXT:    [[TMP24:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP25]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL13:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST12]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST11]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST12]], ptr align 4 [[CALL13]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP25]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT14]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST12]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT15]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST11]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE16:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT14]], [[TMP23]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE16]], label [[OMP_ARRAYCPY_DONE17]], label [[OMP_ARRAYCPY_BODY10]]
// CHECK1:       omp.arraycpy.done17:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP20]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN18:%.*]] = getelementptr inbounds [5 x %struct.S], ptr [[VVAR22]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN18]], i64 5
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP26]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN18]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE19:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done19:
// CHECK1-NEXT:    [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP28]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.8.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[TMP7]], i64 5
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.9
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(16) [[VAR3:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR3_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VAR34:%.*]] = alloca [2 x %struct.S], align 4
// CHECK1-NEXT:    [[_TMP5:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR3]], ptr [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [4 x %struct.S], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw [4 x %struct.S], ptr [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[VAR34]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = ptrtoint ptr [[TMP5]] to i64
// CHECK1-NEXT:    [[TMP7:%.*]] = ptrtoint ptr [[ARRAYIDX]] to i64
// CHECK1-NEXT:    [[TMP8:%.*]] = sub i64 [[TMP6]], [[TMP7]]
// CHECK1-NEXT:    [[TMP9:%.*]] = sdiv exact i64 [[TMP8]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr [[STRUCT_S]], ptr [[VAR34]], i64 [[TMP9]]
// CHECK1-NEXT:    store ptr [[TMP10]], ptr [[_TMP5]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP12]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP13]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1
// CHECK1-NEXT:    store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])
// CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VAR34]], ptr [[TMP22]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
// CHECK1-NEXT:    [[TMP25:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP24]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.9.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP25]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX]], [[TMP26]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE11:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VAR34]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST8:%.*]] = phi ptr [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT9:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST8]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST8]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT9]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST8]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE10:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT9]], [[TMP26]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_DONE11]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done11:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP24]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY12:%.*]] = icmp eq ptr [[ARRAYIDX]], [[TMP27]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY12]], label [[OMP_ARRAYCPY_DONE20:%.*]], label [[OMP_ARRAYCPY_BODY13:%.*]]
// CHECK1:       omp.arraycpy.body13:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST14:%.*]] = phi ptr [ [[VAR34]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT18:%.*]], [[OMP_ARRAYCPY_BODY13]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST15:%.*]] = phi ptr [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT17:%.*]], [[OMP_ARRAYCPY_BODY13]] ]
// CHECK1-NEXT:    [[TMP28:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP29:%.*]] = load i32, ptr [[TMP28]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP29]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL16:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST15]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST14]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST15]], ptr align 4 [[CALL16]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP29]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT17]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT18]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST14]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE19:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT17]], [[TMP27]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE19]], label [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_BODY13]]
// CHECK1:       omp.arraycpy.done20:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP24]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN21:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[VAR34]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN21]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP30]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN21]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE22:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done22:
// CHECK1-NEXT:    [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP32]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.9.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[TMP7]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.10
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(16) [[VAR3:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR3_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VAR34:%.*]] = alloca [2 x %struct.S], align 4
// CHECK1-NEXT:    [[_TMP5:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR3]], ptr [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [4 x %struct.S], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw [4 x %struct.S], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[VAR34]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[ARRAY_BEGIN]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP4]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP4]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = ptrtoint ptr [[TMP5]] to i64
// CHECK1-NEXT:    [[TMP7:%.*]] = ptrtoint ptr [[ARRAYIDX]] to i64
// CHECK1-NEXT:    [[TMP8:%.*]] = sub i64 [[TMP6]], [[TMP7]]
// CHECK1-NEXT:    [[TMP9:%.*]] = sdiv exact i64 [[TMP8]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr [[STRUCT_S]], ptr [[VAR34]], i64 [[TMP9]]
// CHECK1-NEXT:    store ptr [[TMP10]], ptr [[_TMP5]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[TMP11]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP12]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP13]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP14]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP15]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP6:%.*]] = icmp sle i32 [[TMP16]], [[TMP17]]
// CHECK1-NEXT:    br i1 [[CMP6]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP18]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP19]], 1
// CHECK1-NEXT:    store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP20:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[TMP20]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP21]])
// CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VAR34]], ptr [[TMP22]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, ptr [[TMP23]], align 4
// CHECK1-NEXT:    [[TMP25:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP24]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.10.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP25]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP26:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX]], [[TMP26]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE11:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VAR34]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST8:%.*]] = phi ptr [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT9:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST8]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST8]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT9]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST8]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE10:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT9]], [[TMP26]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_DONE11]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done11:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP24]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY12:%.*]] = icmp eq ptr [[ARRAYIDX]], [[TMP27]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY12]], label [[OMP_ARRAYCPY_DONE20:%.*]], label [[OMP_ARRAYCPY_BODY13:%.*]]
// CHECK1:       omp.arraycpy.body13:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST14:%.*]] = phi ptr [ [[VAR34]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT18:%.*]], [[OMP_ARRAYCPY_BODY13]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST15:%.*]] = phi ptr [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT17:%.*]], [[OMP_ARRAYCPY_BODY13]] ]
// CHECK1-NEXT:    [[TMP28:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP29:%.*]] = load i32, ptr [[TMP28]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP29]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL16:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST15]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST14]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST15]], ptr align 4 [[CALL16]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP29]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT17]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST15]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT18]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST14]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE19:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT17]], [[TMP27]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE19]], label [[OMP_ARRAYCPY_DONE20]], label [[OMP_ARRAYCPY_BODY13]]
// CHECK1:       omp.arraycpy.done20:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP24]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN21:%.*]] = getelementptr inbounds [2 x %struct.S], ptr [[VAR34]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN21]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP30]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN21]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE22:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done22:
// CHECK1-NEXT:    [[TMP31:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP32:%.*]] = load i32, ptr [[TMP31]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP32]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.10.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[TMP7]], i64 2
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.11
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(16) [[VAR3:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR3_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[SAVED_STACK:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[__VLA_EXPR0:%.*]] = alloca i64, align 8
// CHECK1-NEXT:    [[_TMP4:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR3]], ptr [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [4 x %struct.S], ptr [[TMP2]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw [4 x %struct.S], ptr [[TMP3]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP4:%.*]] = ptrtoint ptr [[ARRAYIDX3]] to i64
// CHECK1-NEXT:    [[TMP5:%.*]] = ptrtoint ptr [[ARRAYIDX]] to i64
// CHECK1-NEXT:    [[TMP6:%.*]] = sub i64 [[TMP4]], [[TMP5]]
// CHECK1-NEXT:    [[TMP7:%.*]] = sdiv exact i64 [[TMP6]], ptrtoint (ptr getelementptr ([[STRUCT_S:%.*]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP8:%.*]] = add nuw i64 [[TMP7]], 1
// CHECK1-NEXT:    [[TMP9:%.*]] = mul nuw i64 [[TMP8]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP10:%.*]] = call ptr @llvm.stacksave.p0()
// CHECK1-NEXT:    store ptr [[TMP10]], ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT:    [[VLA:%.*]] = alloca [[STRUCT_S]], i64 [[TMP8]], align 16
// CHECK1-NEXT:    store i64 [[TMP8]], ptr [[__VLA_EXPR0]], align 8
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr [[STRUCT_S]], ptr [[VLA]], i64 [[TMP8]]
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[VLA]], [[TMP11]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP11]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = ptrtoint ptr [[TMP12]] to i64
// CHECK1-NEXT:    [[TMP14:%.*]] = ptrtoint ptr [[ARRAYIDX]] to i64
// CHECK1-NEXT:    [[TMP15:%.*]] = sub i64 [[TMP13]], [[TMP14]]
// CHECK1-NEXT:    [[TMP16:%.*]] = sdiv exact i64 [[TMP15]], ptrtoint (ptr getelementptr ([[STRUCT_S]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP17:%.*]] = getelementptr [[STRUCT_S]], ptr [[VLA]], i64 [[TMP16]]
// CHECK1-NEXT:    store ptr [[TMP17]], ptr [[_TMP4]], align 8
// CHECK1-NEXT:    [[TMP18:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[TMP18]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP19]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP20]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP21]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP22]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP23]], [[TMP24]]
// CHECK1-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP25]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD6:%.*]] = add nsw i32 [[TMP26]], 1
// CHECK1-NEXT:    store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP27:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = load i32, ptr [[TMP27]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP28]])
// CHECK1-NEXT:    [[TMP29:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VLA]], ptr [[TMP29]], align 8
// CHECK1-NEXT:    [[TMP30:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP31:%.*]] = inttoptr i64 [[TMP8]] to ptr
// CHECK1-NEXT:    store ptr [[TMP31]], ptr [[TMP30]], align 8
// CHECK1-NEXT:    [[TMP32:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP33:%.*]] = load i32, ptr [[TMP32]], align 4
// CHECK1-NEXT:    [[TMP34:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP33]], i32 1, i64 16, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.11.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP34]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP35:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX]], i64 [[TMP8]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX]], [[TMP35]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE10:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST7:%.*]] = phi ptr [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST7]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST7]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST7]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE9:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP35]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done10:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP33]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP36:%.*]] = getelementptr [[STRUCT_S]], ptr [[ARRAYIDX]], i64 [[TMP8]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY11:%.*]] = icmp eq ptr [[ARRAYIDX]], [[TMP36]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY11]], label [[OMP_ARRAYCPY_DONE19:%.*]], label [[OMP_ARRAYCPY_BODY12:%.*]]
// CHECK1:       omp.arraycpy.body12:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST13:%.*]] = phi ptr [ [[VLA]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT17:%.*]], [[OMP_ARRAYCPY_BODY12]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST14:%.*]] = phi ptr [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT16:%.*]], [[OMP_ARRAYCPY_BODY12]] ]
// CHECK1-NEXT:    [[TMP37:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP38:%.*]] = load i32, ptr [[TMP37]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP38]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL15:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST14]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST13]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST14]], ptr align 4 [[CALL15]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP38]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT16]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST14]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT17]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST13]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE18:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT16]], [[TMP36]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_DONE19]], label [[OMP_ARRAYCPY_BODY12]]
// CHECK1:       omp.arraycpy.done19:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP33]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[TMP39:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[VLA]], i64 [[TMP8]]
// CHECK1-NEXT:    [[ARRAYDESTROY_ISEMPTY:%.*]] = icmp eq ptr [[VLA]], [[TMP39]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_ISEMPTY]], label [[ARRAYDESTROY_DONE20:%.*]], label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP39]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[VLA]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE20]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done20:
// CHECK1-NEXT:    [[TMP40:%.*]] = load ptr, ptr [[SAVED_STACK]], align 8
// CHECK1-NEXT:    call void @llvm.stackrestore.p0(ptr [[TMP40]])
// CHECK1-NEXT:    [[TMP41:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP42:%.*]] = load i32, ptr [[TMP41]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP42]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.11.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = ptrtoint ptr [[TMP9]] to i64
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[TMP7]], i64 [[TMP10]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP11]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP11]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.12
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(16) [[VAR3:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR3_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VAR33:%.*]] = alloca [4 x %struct.S], align 16
// CHECK1-NEXT:    [[_TMP4:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR3]], ptr [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VAR3_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 9, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [4 x %struct.S], ptr [[VAR33]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[ARRAY_BEGIN]], i64 4
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP3]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @_ZN1SIfEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP3]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    store ptr [[VAR33]], ptr [[_TMP4]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[TMP4]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP5]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP6]], 9
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 9, [[COND_TRUE]] ], [ [[TMP7]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP8]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP9]], [[TMP10]]
// CHECK1-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP11]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD6:%.*]] = add nsw i32 [[TMP12]], 1
// CHECK1-NEXT:    store i32 [[ADD6]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP13:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[TMP13]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP14]])
// CHECK1-NEXT:    [[TMP15:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[VAR33]], ptr [[TMP15]], align 8
// CHECK1-NEXT:    [[TMP16:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[TMP16]], align 4
// CHECK1-NEXT:    [[TMP18:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP17]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @main.omp_outlined.12.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP18]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP19:%.*]] = getelementptr [[STRUCT_S]], ptr [[TMP2]], i64 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP2]], [[TMP19]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE10:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[VAR33]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST7:%.*]] = phi ptr [ [[TMP2]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT8:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST7]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST7]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT8]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST7]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE9:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT8]], [[TMP19]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE9]], label [[OMP_ARRAYCPY_DONE10]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done10:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP17]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP20:%.*]] = getelementptr [[STRUCT_S]], ptr [[TMP2]], i64 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY11:%.*]] = icmp eq ptr [[TMP2]], [[TMP20]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY11]], label [[OMP_ARRAYCPY_DONE19:%.*]], label [[OMP_ARRAYCPY_BODY12:%.*]]
// CHECK1:       omp.arraycpy.body12:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST13:%.*]] = phi ptr [ [[VAR33]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT17:%.*]], [[OMP_ARRAYCPY_BODY12]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST14:%.*]] = phi ptr [ [[TMP2]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT16:%.*]], [[OMP_ARRAYCPY_BODY12]] ]
// CHECK1-NEXT:    [[TMP21:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP22:%.*]] = load i32, ptr [[TMP21]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP22]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL15:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST14]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST13]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST14]], ptr align 4 [[CALL15]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP22]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT16]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST14]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT17]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST13]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE18:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT16]], [[TMP20]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE18]], label [[OMP_ARRAYCPY_DONE19]], label [[OMP_ARRAYCPY_BODY12]]
// CHECK1:       omp.arraycpy.done19:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP17]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN20:%.*]] = getelementptr inbounds [4 x %struct.S], ptr [[VAR33]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAY_BEGIN20]], i64 4
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP23]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIfED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN20]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE21:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done21:
// CHECK1-NEXT:    [[TMP24:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, ptr [[TMP24]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP25]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@main.omp_outlined.12.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S:%.*]], ptr [[TMP7]], i64 4
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE2:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIfEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE2]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done2:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v
// CHECK1-SAME: () #[[ATTR1]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[TEST:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[T_VAR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T_VAR1:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VEC:%.*]] = alloca [2 x i32], align 4
// CHECK1-NEXT:    [[S_ARR:%.*]] = alloca [2 x %struct.S.0], align 4
// CHECK1-NEXT:    [[VAR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR1:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK1-NEXT:    [[ARR:%.*]] = alloca [42 x %struct.S.0], align 16
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[TEST]])
// CHECK1-NEXT:    store i32 0, ptr [[T_VAR]], align 4
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[VEC]], ptr align 4 @__const._Z5tmainIiLi42EET_v.vec, i64 8, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[S_ARR]], i32 noundef 1)
// CHECK1-NEXT:    [[ARRAYINIT_ELEMENT:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[S_ARR]], i64 1
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYINIT_ELEMENT]], i32 noundef 2)
// CHECK1-NEXT:    store ptr [[TEST]], ptr [[VAR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR1]])
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [42 x %struct.S.0], ptr [[ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[ARRAYCTOR_END:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i64 42
// CHECK1-NEXT:    br label [[ARRAYCTOR_LOOP:%.*]]
// CHECK1:       arrayctor.loop:
// CHECK1-NEXT:    [[ARRAYCTOR_CUR:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[ARRAYCTOR_NEXT:%.*]], [[ARRAYCTOR_LOOP]] ]
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYCTOR_CUR]])
// CHECK1-NEXT:    [[ARRAYCTOR_NEXT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYCTOR_CUR]], i64 1
// CHECK1-NEXT:    [[ARRAYCTOR_DONE:%.*]] = icmp eq ptr [[ARRAYCTOR_NEXT]], [[ARRAYCTOR_END]]
// CHECK1-NEXT:    br i1 [[ARRAYCTOR_DONE]], label [[ARRAYCTOR_CONT:%.*]], label [[ARRAYCTOR_LOOP]]
// CHECK1:       arrayctor.cont:
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[VAR]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 6, ptr @_Z5tmainIiLi42EET_v.omp_outlined, ptr [[T_VAR]], ptr [[TMP0]], ptr [[VAR1]], ptr [[T_VAR1]], ptr [[VEC]], ptr [[S_ARR]])
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[VAR]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 4, ptr @_Z5tmainIiLi42EET_v.omp_outlined.13, ptr [[T_VAR]], ptr [[VEC]], ptr [[S_ARR]], ptr [[TMP1]])
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[VAR]], align 8
// CHECK1-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB3]], i32 5, ptr @_Z5tmainIiLi42EET_v.omp_outlined.14, ptr [[ARR]], ptr [[VEC]], ptr [[T_VAR]], ptr [[S_ARR]], ptr [[TMP2]])
// CHECK1-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    [[ARRAY_BEGIN1:%.*]] = getelementptr inbounds [42 x %struct.S.0], ptr [[ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP3:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN1]], i64 42
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP3]], [[ARRAYCTOR_CONT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN1]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE2:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done2:
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR1]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAY_BEGIN3:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[S_ARR]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN3]], i64 2
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY4:%.*]]
// CHECK1:       arraydestroy.body4:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST5:%.*]] = phi ptr [ [[TMP4]], [[ARRAYDESTROY_DONE2]] ], [ [[ARRAYDESTROY_ELEMENT6:%.*]], [[ARRAYDESTROY_BODY4]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT6]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST5]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT6]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE7:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT6]], [[ARRAY_BEGIN3]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE7]], label [[ARRAYDESTROY_DONE8:%.*]], label [[ARRAYDESTROY_BODY4]]
// CHECK1:       arraydestroy.done8:
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[TEST]]) #[[ATTR5]]
// CHECK1-NEXT:    [[TMP5:%.*]] = load i32, ptr [[RETVAL]], align 4
// CHECK1-NEXT:    ret i32 [[TMP5]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load volatile double, ptr @g, align 8
// CHECK1-NEXT:    [[CONV:%.*]] = fptrunc double [[TMP0]] to float
// CHECK1-NEXT:    store float [[CONV]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfEC2Ef
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], float noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca float, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store float [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load float, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[CONV:%.*]] = fpext float [[TMP0]] to double
// CHECK1-NEXT:    [[TMP1:%.*]] = load volatile double, ptr @g, align 8
// CHECK1-NEXT:    [[ADD:%.*]] = fadd double [[CONV]], [[TMP1]]
// CHECK1-NEXT:    [[CONV2:%.*]] = fptrunc double [[ADD]] to float
// CHECK1-NEXT:    store float [[CONV2]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIfED2Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiEC2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC1Ei
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    call void @_ZN1SIiEC2Ei(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]], i32 noundef [[TMP0]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v.omp_outlined
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR1:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR1:%.*]], ptr noundef nonnull align 4 dereferenceable(8) [[VEC:%.*]], ptr noundef nonnull align 4 dereferenceable(8) [[S_ARR:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR1_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[T_VAR1_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T_VAR3:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[VAR4:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[_TMP5:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR16:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK1-NEXT:    [[T_VAR17:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [4 x ptr], align 8
// CHECK1-NEXT:    [[REF_TMP:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK1-NEXT:    [[REF_TMP22:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[T_VAR]], ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR]], ptr [[VAR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR1]], ptr [[VAR1_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[T_VAR1]], ptr [[T_VAR1_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VEC]], ptr [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[S_ARR]], ptr [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[VAR1_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[T_VAR1_ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[TMP]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT:    store ptr [[TMP6]], ptr [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR4]])
// CHECK1-NEXT:    store ptr [[VAR4]], ptr [[_TMP5]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR16]])
// CHECK1-NEXT:    store i32 2147483647, ptr [[T_VAR17]], align 4
// CHECK1-NEXT:    [[TMP8:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[TMP8]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP9]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP10]], 1
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP11]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP12]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP8:%.*]] = icmp sle i32 [[TMP13]], [[TMP14]]
// CHECK1-NEXT:    br i1 [[CMP8]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP15]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP17]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[TMP4]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 [[TMP16]], ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[TMP18:%.*]] = load ptr, ptr [[_TMP5]], align 8
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM9:%.*]] = sext i32 [[TMP19]] to i64
// CHECK1-NEXT:    [[ARRAYIDX10:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[TMP5]], i64 0, i64 [[IDXPROM9]]
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[ARRAYIDX10]], ptr align 4 [[TMP18]], i64 4, i1 false)
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD11:%.*]] = add nsw i32 [[TMP20]], 1
// CHECK1-NEXT:    store i32 [[ADD11]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP9]])
// CHECK1-NEXT:    [[TMP21:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[T_VAR3]], ptr [[TMP21]], align 8
// CHECK1-NEXT:    [[TMP22:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK1-NEXT:    store ptr [[VAR4]], ptr [[TMP22]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 2
// CHECK1-NEXT:    store ptr [[VAR16]], ptr [[TMP23]], align 8
// CHECK1-NEXT:    [[TMP24:%.*]] = getelementptr inbounds [4 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 3
// CHECK1-NEXT:    store ptr [[T_VAR17]], ptr [[TMP24]], align 8
// CHECK1-NEXT:    [[TMP25:%.*]] = call i32 @__kmpc_reduce_nowait(ptr @[[GLOB2]], i32 [[TMP9]], i32 4, i64 32, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @_Z5tmainIiLi42EET_v.omp_outlined.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP25]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT:    [[TMP27:%.*]] = load i32, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[ADD12:%.*]] = add nsw i32 [[TMP26]], [[TMP27]]
// CHECK1-NEXT:    store i32 [[ADD12]], ptr [[TMP0]], align 4
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIiEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[TMP7]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR4]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP7]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[CALL13:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK1-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[CALL13]], 0
// CHECK1-NEXT:    br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]]
// CHECK1:       land.rhs:
// CHECK1-NEXT:    [[CALL14:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[VAR16]])
// CHECK1-NEXT:    [[TOBOOL15:%.*]] = icmp ne i32 [[CALL14]], 0
// CHECK1-NEXT:    br label [[LAND_END]]
// CHECK1:       land.end:
// CHECK1-NEXT:    [[TMP28:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE1]] ], [ [[TOBOOL15]], [[LAND_RHS]] ]
// CHECK1-NEXT:    [[CONV:%.*]] = zext i1 [[TMP28]] to i32
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]], i32 noundef [[CONV]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP2]], ptr align 4 [[REF_TMP]], i64 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]]
// CHECK1-NEXT:    [[TMP29:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT:    [[TMP30:%.*]] = load i32, ptr [[T_VAR17]], align 4
// CHECK1-NEXT:    [[CMP16:%.*]] = icmp slt i32 [[TMP29]], [[TMP30]]
// CHECK1-NEXT:    br i1 [[CMP16]], label [[COND_TRUE17:%.*]], label [[COND_FALSE18:%.*]]
// CHECK1:       cond.true17:
// CHECK1-NEXT:    [[TMP31:%.*]] = load i32, ptr [[TMP3]], align 4
// CHECK1-NEXT:    br label [[COND_END19:%.*]]
// CHECK1:       cond.false18:
// CHECK1-NEXT:    [[TMP32:%.*]] = load i32, ptr [[T_VAR17]], align 4
// CHECK1-NEXT:    br label [[COND_END19]]
// CHECK1:       cond.end19:
// CHECK1-NEXT:    [[COND20:%.*]] = phi i32 [ [[TMP31]], [[COND_TRUE17]] ], [ [[TMP32]], [[COND_FALSE18]] ]
// CHECK1-NEXT:    store i32 [[COND20]], ptr [[TMP3]], align 4
// CHECK1-NEXT:    call void @__kmpc_end_reduce_nowait(ptr @[[GLOB2]], i32 [[TMP9]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP33:%.*]] = load i32, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TMP34:%.*]] = atomicrmw add ptr [[TMP0]], i32 [[TMP33]] monotonic, align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP9]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL21:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIiEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[TMP7]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR4]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP7]], ptr align 4 [[CALL21]], i64 4, i1 false)
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP9]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP9]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL23:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[TMP2]])
// CHECK1-NEXT:    [[TOBOOL24:%.*]] = icmp ne i32 [[CALL23]], 0
// CHECK1-NEXT:    br i1 [[TOBOOL24]], label [[LAND_RHS25:%.*]], label [[LAND_END28:%.*]]
// CHECK1:       land.rhs25:
// CHECK1-NEXT:    [[CALL26:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[VAR16]])
// CHECK1-NEXT:    [[TOBOOL27:%.*]] = icmp ne i32 [[CALL26]], 0
// CHECK1-NEXT:    br label [[LAND_END28]]
// CHECK1:       land.end28:
// CHECK1-NEXT:    [[TMP35:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE2]] ], [ [[TOBOOL27]], [[LAND_RHS25]] ]
// CHECK1-NEXT:    [[CONV29:%.*]] = zext i1 [[TMP35]] to i32
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP22]], i32 noundef [[CONV29]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP2]], ptr align 4 [[REF_TMP22]], i64 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP22]]) #[[ATTR5]]
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP9]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[TMP36:%.*]] = load i32, ptr [[T_VAR17]], align 4
// CHECK1-NEXT:    [[TMP37:%.*]] = atomicrmw min ptr [[TMP3]], i32 [[TMP36]] monotonic, align 4
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR16]]) #[[ATTR5]]
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[VAR4]]) #[[ATTR5]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v.omp_outlined.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[REF_TMP:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP3]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK1-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK1-NEXT:    [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
// CHECK1-NEXT:    [[TMP12:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP3]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP13:%.*]] = load ptr, ptr [[TMP12]], align 8
// CHECK1-NEXT:    [[TMP14:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 2
// CHECK1-NEXT:    [[TMP15:%.*]] = load ptr, ptr [[TMP14]], align 8
// CHECK1-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP3]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP17:%.*]] = load ptr, ptr [[TMP16]], align 8
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [4 x ptr], ptr [[TMP2]], i64 0, i64 3
// CHECK1-NEXT:    [[TMP19:%.*]] = load ptr, ptr [[TMP18]], align 8
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[TMP20]], [[TMP21]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[TMP7]], align 4
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef nonnull align 4 dereferenceable(4) ptr @_ZN1SIiEanERKS0_(ptr noundef nonnull align 4 dereferenceable(4) [[TMP11]], ptr noundef nonnull align 4 dereferenceable(4) [[TMP9]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP11]], ptr align 4 [[CALL]], i64 4, i1 false)
// CHECK1-NEXT:    [[CALL2:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[TMP15]])
// CHECK1-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[CALL2]], 0
// CHECK1-NEXT:    br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]]
// CHECK1:       land.rhs:
// CHECK1-NEXT:    [[CALL3:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[TMP13]])
// CHECK1-NEXT:    [[TOBOOL4:%.*]] = icmp ne i32 [[CALL3]], 0
// CHECK1-NEXT:    br label [[LAND_END]]
// CHECK1:       land.end:
// CHECK1-NEXT:    [[TMP22:%.*]] = phi i1 [ false, [[ENTRY:%.*]] ], [ [[TOBOOL4]], [[LAND_RHS]] ]
// CHECK1-NEXT:    [[CONV:%.*]] = zext i1 [[TMP22]] to i32
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]], i32 noundef [[CONV]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[TMP15]], ptr align 4 [[REF_TMP]], i64 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]]
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, ptr [[TMP19]], align 4
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, ptr [[TMP17]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp slt i32 [[TMP23]], [[TMP24]]
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, ptr [[TMP19]], align 4
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, ptr [[TMP17]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ [[TMP25]], [[COND_TRUE]] ], [ [[TMP26]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[TMP19]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEanERKS0_
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[TMP0:%.*]]) #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    ret ptr [[THIS1]]
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEcviEv
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    ret i32 0
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED1Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    call void @_ZN1SIiED2Ev(ptr noundef nonnull align 4 dereferenceable(4) [[THIS1]]) #[[ATTR5]]
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v.omp_outlined.13
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], ptr noundef nonnull align 4 dereferenceable(8) [[VEC:%.*]], ptr noundef nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[T_VAR3:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[ATOMIC_TEMP:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP10:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[T_VAR]], ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VEC]], ptr [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[S_ARR]], ptr [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR]], ptr [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[VAR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP3]], ptr [[TMP]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT:    store ptr [[TMP4]], ptr [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP6]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP7]], 1
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP8]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP9]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP10:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP11:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP4:%.*]] = icmp sle i32 [[TMP10]], [[TMP11]]
// CHECK1-NEXT:    br i1 [[CMP4]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP12:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP12]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP14]] to i64
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds [2 x i32], ptr [[TMP1]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 [[TMP13]], ptr [[ARRAYIDX]], align 4
// CHECK1-NEXT:    [[TMP15:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM5:%.*]] = sext i32 [[TMP16]] to i64
// CHECK1-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[TMP2]], i64 0, i64 [[IDXPROM5]]
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[ARRAYIDX6]], ptr align 4 [[TMP15]], i64 4, i1 false)
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD7:%.*]] = add nsw i32 [[TMP17]], 1
// CHECK1-NEXT:    store i32 [[ADD7]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP6]])
// CHECK1-NEXT:    [[TMP18:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[T_VAR3]], ptr [[TMP18]], align 8
// CHECK1-NEXT:    [[TMP19:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP6]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @_Z5tmainIiLi42EET_v.omp_outlined.13.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP19]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[TMP0]], align 4
// CHECK1-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP20]], 0
// CHECK1-NEXT:    br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]]
// CHECK1:       land.rhs:
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TOBOOL8:%.*]] = icmp ne i32 [[TMP21]], 0
// CHECK1-NEXT:    br label [[LAND_END]]
// CHECK1:       land.end:
// CHECK1-NEXT:    [[TMP22:%.*]] = phi i1 [ false, [[DOTOMP_REDUCTION_CASE1]] ], [ [[TOBOOL8]], [[LAND_RHS]] ]
// CHECK1-NEXT:    [[CONV:%.*]] = zext i1 [[TMP22]] to i32
// CHECK1-NEXT:    store i32 [[CONV]], ptr [[TMP0]], align 4
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP6]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TOBOOL9:%.*]] = icmp ne i32 [[TMP23]], 0
// CHECK1-NEXT:    [[ATOMIC_LOAD:%.*]] = load atomic i32, ptr [[TMP0]] monotonic, align 4
// CHECK1-NEXT:    br label [[ATOMIC_CONT:%.*]]
// CHECK1:       atomic_cont:
// CHECK1-NEXT:    [[TMP24:%.*]] = phi i32 [ [[ATOMIC_LOAD]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[TMP30:%.*]], [[LAND_END14:%.*]] ]
// CHECK1-NEXT:    store i32 [[TMP24]], ptr [[_TMP10]], align 4
// CHECK1-NEXT:    [[TMP25:%.*]] = load i32, ptr [[_TMP10]], align 4
// CHECK1-NEXT:    [[TOBOOL11:%.*]] = icmp ne i32 [[TMP25]], 0
// CHECK1-NEXT:    br i1 [[TOBOOL11]], label [[LAND_RHS12:%.*]], label [[LAND_END14]]
// CHECK1:       land.rhs12:
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, ptr [[T_VAR3]], align 4
// CHECK1-NEXT:    [[TOBOOL13:%.*]] = icmp ne i32 [[TMP26]], 0
// CHECK1-NEXT:    br label [[LAND_END14]]
// CHECK1:       land.end14:
// CHECK1-NEXT:    [[TMP27:%.*]] = phi i1 [ false, [[ATOMIC_CONT]] ], [ [[TOBOOL13]], [[LAND_RHS12]] ]
// CHECK1-NEXT:    [[CONV15:%.*]] = zext i1 [[TMP27]] to i32
// CHECK1-NEXT:    store i32 [[CONV15]], ptr [[ATOMIC_TEMP]], align 4
// CHECK1-NEXT:    [[TMP28:%.*]] = load i32, ptr [[ATOMIC_TEMP]], align 4
// CHECK1-NEXT:    [[TMP29:%.*]] = cmpxchg ptr [[TMP0]], i32 [[TMP24]], i32 [[TMP28]] monotonic monotonic, align 4
// CHECK1-NEXT:    [[TMP30]] = extractvalue { i32, i1 } [[TMP29]], 0
// CHECK1-NEXT:    [[TMP31:%.*]] = extractvalue { i32, i1 } [[TMP29]], 1
// CHECK1-NEXT:    br i1 [[TMP31]], label [[ATOMIC_EXIT:%.*]], label [[ATOMIC_CONT]]
// CHECK1:       atomic_exit:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP6]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP6]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v.omp_outlined.13.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = load i32, ptr [[TMP7]], align 4
// CHECK1-NEXT:    [[TOBOOL:%.*]] = icmp ne i32 [[TMP8]], 0
// CHECK1-NEXT:    br i1 [[TOBOOL]], label [[LAND_RHS:%.*]], label [[LAND_END:%.*]]
// CHECK1:       land.rhs:
// CHECK1-NEXT:    [[TMP9:%.*]] = load i32, ptr [[TMP5]], align 4
// CHECK1-NEXT:    [[TOBOOL2:%.*]] = icmp ne i32 [[TMP9]], 0
// CHECK1-NEXT:    br label [[LAND_END]]
// CHECK1:       land.end:
// CHECK1-NEXT:    [[TMP10:%.*]] = phi i1 [ false, [[ENTRY:%.*]] ], [ [[TOBOOL2]], [[LAND_RHS]] ]
// CHECK1-NEXT:    [[CONV:%.*]] = zext i1 [[TMP10]] to i32
// CHECK1-NEXT:    store i32 [[CONV]], ptr [[TMP7]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v.omp_outlined.14
// CHECK1-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]], ptr noundef nonnull align 4 dereferenceable(168) [[ARR:%.*]], ptr noundef nonnull align 4 dereferenceable(8) [[VEC:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[T_VAR:%.*]], ptr noundef nonnull align 4 dereferenceable(8) [[S_ARR:%.*]], ptr noundef nonnull align 4 dereferenceable(4) [[VAR:%.*]]) #[[ATTR4]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VEC_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[T_VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[S_ARR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[VAR_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[_TMP1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[_TMP2:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[ARR4:%.*]] = alloca [40 x %struct.S.0], align 16
// CHECK1-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [1 x ptr], align 8
// CHECK1-NEXT:    [[REF_TMP:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    [[REF_TMP20:%.*]] = alloca [[STRUCT_S_0]], align 4
// CHECK1-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK1-NEXT:    store ptr [[ARR]], ptr [[ARR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VEC]], ptr [[VEC_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[T_VAR]], ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[S_ARR]], ptr [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[VAR]], ptr [[VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP1:%.*]] = load ptr, ptr [[VEC_ADDR]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[T_VAR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[S_ARR_ADDR]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = load ptr, ptr [[VAR_ADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP4]], ptr [[TMP]], align 8
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP]], align 8
// CHECK1-NEXT:    store ptr [[TMP5]], ptr [[_TMP1]], align 8
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK1-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK1-NEXT:    [[ARRAYIDX:%.*]] = getelementptr inbounds nuw [42 x %struct.S.0], ptr [[TMP0]], i64 0, i64 1
// CHECK1-NEXT:    [[ARRAYIDX3:%.*]] = getelementptr inbounds nuw [42 x %struct.S.0], ptr [[TMP0]], i64 0, i64 40
// CHECK1-NEXT:    [[ARRAY_BEGIN:%.*]] = getelementptr inbounds [40 x %struct.S.0], ptr [[ARR4]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr [[STRUCT_S_0]], ptr [[ARRAY_BEGIN]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYINIT_ISEMPTY:%.*]] = icmp eq ptr [[ARRAY_BEGIN]], [[TMP6]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYINIT_ISEMPTY]], label [[OMP_ARRAYINIT_DONE:%.*]], label [[OMP_ARRAYINIT_BODY:%.*]]
// CHECK1:       omp.arrayinit.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[ARRAY_BEGIN]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYINIT_BODY]] ]
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]])
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP6]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYINIT_DONE]], label [[OMP_ARRAYINIT_BODY]]
// CHECK1:       omp.arrayinit.done:
// CHECK1-NEXT:    [[TMP7:%.*]] = ptrtoint ptr [[TMP0]] to i64
// CHECK1-NEXT:    [[TMP8:%.*]] = ptrtoint ptr [[ARRAYIDX]] to i64
// CHECK1-NEXT:    [[TMP9:%.*]] = sub i64 [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    [[TMP10:%.*]] = sdiv exact i64 [[TMP9]], ptrtoint (ptr getelementptr ([[STRUCT_S_0]], ptr null, i32 1) to i64)
// CHECK1-NEXT:    [[TMP11:%.*]] = getelementptr [[STRUCT_S_0]], ptr [[ARR4]], i64 [[TMP10]]
// CHECK1-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP13:%.*]] = load i32, ptr [[TMP12]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1]], i32 [[TMP13]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK1-NEXT:    [[TMP14:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP14]], 1
// CHECK1-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK1:       cond.true:
// CHECK1-NEXT:    br label [[COND_END:%.*]]
// CHECK1:       cond.false:
// CHECK1-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    br label [[COND_END]]
// CHECK1:       cond.end:
// CHECK1-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP15]], [[COND_FALSE]] ]
// CHECK1-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[TMP16:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK1-NEXT:    store i32 [[TMP16]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK1:       omp.inner.for.cond:
// CHECK1-NEXT:    [[TMP17:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[TMP18:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK1-NEXT:    [[CMP5:%.*]] = icmp sle i32 [[TMP17]], [[TMP18]]
// CHECK1-NEXT:    br i1 [[CMP5]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_COND_CLEANUP:%.*]]
// CHECK1:       omp.inner.for.cond.cleanup:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_END:%.*]]
// CHECK1:       omp.inner.for.body:
// CHECK1-NEXT:    [[TMP19:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP19]], 1
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK1-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK1-NEXT:    [[TMP20:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK1-NEXT:    [[TMP21:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM:%.*]] = sext i32 [[TMP21]] to i64
// CHECK1-NEXT:    [[ARRAYIDX6:%.*]] = getelementptr inbounds [2 x i32], ptr [[TMP1]], i64 0, i64 [[IDXPROM]]
// CHECK1-NEXT:    store i32 [[TMP20]], ptr [[ARRAYIDX6]], align 4
// CHECK1-NEXT:    [[TMP22:%.*]] = load ptr, ptr [[_TMP1]], align 8
// CHECK1-NEXT:    [[TMP23:%.*]] = load i32, ptr [[I]], align 4
// CHECK1-NEXT:    [[IDXPROM7:%.*]] = sext i32 [[TMP23]] to i64
// CHECK1-NEXT:    [[ARRAYIDX8:%.*]] = getelementptr inbounds [2 x %struct.S.0], ptr [[TMP3]], i64 0, i64 [[IDXPROM7]]
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[ARRAYIDX8]], ptr align 4 [[TMP22]], i64 4, i1 false)
// CHECK1-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK1:       omp.body.continue:
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK1:       omp.inner.for.inc:
// CHECK1-NEXT:    [[TMP24:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    [[ADD9:%.*]] = add nsw i32 [[TMP24]], 1
// CHECK1-NEXT:    store i32 [[ADD9]], ptr [[DOTOMP_IV]], align 4
// CHECK1-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK1:       omp.inner.for.end:
// CHECK1-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK1:       omp.loop.exit:
// CHECK1-NEXT:    [[TMP25:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP26:%.*]] = load i32, ptr [[TMP25]], align 4
// CHECK1-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP26]])
// CHECK1-NEXT:    [[TMP27:%.*]] = getelementptr inbounds [1 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK1-NEXT:    store ptr [[ARR4]], ptr [[TMP27]], align 8
// CHECK1-NEXT:    [[TMP28:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP29:%.*]] = load i32, ptr [[TMP28]], align 4
// CHECK1-NEXT:    [[TMP30:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2]], i32 [[TMP29]], i32 1, i64 8, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @_Z5tmainIiLi42EET_v.omp_outlined.14.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    switch i32 [[TMP30]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK1-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK1-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK1-NEXT:    ]
// CHECK1:       .omp.reduction.case1:
// CHECK1-NEXT:    [[TMP31:%.*]] = getelementptr [[STRUCT_S_0]], ptr [[ARRAYIDX]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[ARRAYIDX]], [[TMP31]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE15:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[ARR4]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST10:%.*]] = phi ptr [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE1]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT13:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST10]])
// CHECK1-NEXT:    [[CALL11:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    [[ADD12:%.*]] = add nsw i32 [[CALL]], [[CALL11]]
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]], i32 noundef [[ADD12]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST10]], ptr align 4 [[REF_TMP]], i64 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT13]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST10]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE14:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT13]], [[TMP31]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE14]], label [[OMP_ARRAYCPY_DONE15]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done15:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP29]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.case2:
// CHECK1-NEXT:    [[TMP32:%.*]] = getelementptr [[STRUCT_S_0]], ptr [[ARRAYIDX]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY16:%.*]] = icmp eq ptr [[ARRAYIDX]], [[TMP32]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY16]], label [[OMP_ARRAYCPY_DONE27:%.*]], label [[OMP_ARRAYCPY_BODY17:%.*]]
// CHECK1:       omp.arraycpy.body17:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST18:%.*]] = phi ptr [ [[ARR4]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT25:%.*]], [[OMP_ARRAYCPY_BODY17]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST19:%.*]] = phi ptr [ [[ARRAYIDX]], [[DOTOMP_REDUCTION_CASE2]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT24:%.*]], [[OMP_ARRAYCPY_BODY17]] ]
// CHECK1-NEXT:    [[TMP33:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP34:%.*]] = load i32, ptr [[TMP33]], align 4
// CHECK1-NEXT:    call void @__kmpc_critical(ptr @[[GLOB3]], i32 [[TMP34]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[CALL21:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST19]])
// CHECK1-NEXT:    [[CALL22:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST18]])
// CHECK1-NEXT:    [[ADD23:%.*]] = add nsw i32 [[CALL21]], [[CALL22]]
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP20]], i32 noundef [[ADD23]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST19]], ptr align 4 [[REF_TMP20]], i64 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP20]]) #[[ATTR5]]
// CHECK1-NEXT:    call void @__kmpc_end_critical(ptr @[[GLOB3]], i32 [[TMP34]], ptr @.gomp_critical_user_.atomic_reduction.var)
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT24]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST19]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT25]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST18]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE26:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT24]], [[TMP32]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE26]], label [[OMP_ARRAYCPY_DONE27]], label [[OMP_ARRAYCPY_BODY17]]
// CHECK1:       omp.arraycpy.done27:
// CHECK1-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP29]], ptr @.gomp_critical_user_.reduction.var)
// CHECK1-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK1:       .omp.reduction.default:
// CHECK1-NEXT:    [[ARRAY_BEGIN28:%.*]] = getelementptr inbounds [40 x %struct.S.0], ptr [[ARR4]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP35:%.*]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAY_BEGIN28]], i64 40
// CHECK1-NEXT:    br label [[ARRAYDESTROY_BODY:%.*]]
// CHECK1:       arraydestroy.body:
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENTPAST:%.*]] = phi ptr [ [[TMP35]], [[DOTOMP_REDUCTION_DEFAULT]] ], [ [[ARRAYDESTROY_ELEMENT:%.*]], [[ARRAYDESTROY_BODY]] ]
// CHECK1-NEXT:    [[ARRAYDESTROY_ELEMENT]] = getelementptr inbounds [[STRUCT_S_0]], ptr [[ARRAYDESTROY_ELEMENTPAST]], i64 -1
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[ARRAYDESTROY_ELEMENT]]) #[[ATTR5]]
// CHECK1-NEXT:    [[ARRAYDESTROY_DONE:%.*]] = icmp eq ptr [[ARRAYDESTROY_ELEMENT]], [[ARRAY_BEGIN28]]
// CHECK1-NEXT:    br i1 [[ARRAYDESTROY_DONE]], label [[ARRAYDESTROY_DONE29:%.*]], label [[ARRAYDESTROY_BODY]]
// CHECK1:       arraydestroy.done29:
// CHECK1-NEXT:    [[TMP36:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK1-NEXT:    [[TMP37:%.*]] = load i32, ptr [[TMP36]], align 4
// CHECK1-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB4]], i32 [[TMP37]])
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_Z5tmainIiLi42EET_v.omp_outlined.14.omp.reduction.reduction_func
// CHECK1-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR6]] {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[REF_TMP:%.*]] = alloca [[STRUCT_S_0:%.*]], align 4
// CHECK1-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK1-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK1-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK1-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [1 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK1-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK1-NEXT:    [[TMP8:%.*]] = getelementptr [[STRUCT_S_0]], ptr [[TMP7]], i64 40
// CHECK1-NEXT:    [[OMP_ARRAYCPY_ISEMPTY:%.*]] = icmp eq ptr [[TMP7]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_ISEMPTY]], label [[OMP_ARRAYCPY_DONE3:%.*]], label [[OMP_ARRAYCPY_BODY:%.*]]
// CHECK1:       omp.arraycpy.body:
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRCELEMENTPAST:%.*]] = phi ptr [ [[TMP5]], [[ENTRY:%.*]] ], [ [[OMP_ARRAYCPY_SRC_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DESTELEMENTPAST:%.*]] = phi ptr [ [[TMP7]], [[ENTRY]] ], [ [[OMP_ARRAYCPY_DEST_ELEMENT:%.*]], [[OMP_ARRAYCPY_BODY]] ]
// CHECK1-NEXT:    [[CALL:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_DESTELEMENTPAST]])
// CHECK1-NEXT:    [[CALL2:%.*]] = call noundef i32 @_ZN1SIiEcviEv(ptr noundef nonnull align 4 dereferenceable(4) [[OMP_ARRAYCPY_SRCELEMENTPAST]])
// CHECK1-NEXT:    [[ADD:%.*]] = add nsw i32 [[CALL]], [[CALL2]]
// CHECK1-NEXT:    call void @_ZN1SIiEC1Ei(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]], i32 noundef [[ADD]])
// CHECK1-NEXT:    call void @llvm.memcpy.p0.p0.i64(ptr align 4 [[OMP_ARRAYCPY_DESTELEMENTPAST]], ptr align 4 [[REF_TMP]], i64 4, i1 false)
// CHECK1-NEXT:    call void @_ZN1SIiED1Ev(ptr noundef nonnull align 4 dereferenceable(4) [[REF_TMP]]) #[[ATTR5]]
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DEST_ELEMENT]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_DESTELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_SRC_ELEMENT]] = getelementptr [[STRUCT_S_0]], ptr [[OMP_ARRAYCPY_SRCELEMENTPAST]], i32 1
// CHECK1-NEXT:    [[OMP_ARRAYCPY_DONE:%.*]] = icmp eq ptr [[OMP_ARRAYCPY_DEST_ELEMENT]], [[TMP8]]
// CHECK1-NEXT:    br i1 [[OMP_ARRAYCPY_DONE]], label [[OMP_ARRAYCPY_DONE3]], label [[OMP_ARRAYCPY_BODY]]
// CHECK1:       omp.arraycpy.done3:
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load volatile double, ptr @g, align 8
// CHECK1-NEXT:    [[CONV:%.*]] = fptosi double [[TMP0]] to i32
// CHECK1-NEXT:    store i32 [[CONV]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiEC2Ei
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]], i32 noundef [[A:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    [[A_ADDR:%.*]] = alloca i32, align 4
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    store i32 [[A]], ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[F:%.*]] = getelementptr inbounds nuw [[STRUCT_S_0:%.*]], ptr [[THIS1]], i32 0, i32 0
// CHECK1-NEXT:    [[TMP0:%.*]] = load i32, ptr [[A_ADDR]], align 4
// CHECK1-NEXT:    [[CONV:%.*]] = sitofp i32 [[TMP0]] to double
// CHECK1-NEXT:    [[TMP1:%.*]] = load volatile double, ptr @g, align 8
// CHECK1-NEXT:    [[ADD:%.*]] = fadd double [[CONV]], [[TMP1]]
// CHECK1-NEXT:    [[CONV2:%.*]] = fptosi double [[ADD]] to i32
// CHECK1-NEXT:    store i32 [[CONV2]], ptr [[F]], align 4
// CHECK1-NEXT:    ret void
//
//
// CHECK1-LABEL: define {{[^@]+}}@_ZN1SIiED2Ev
// CHECK1-SAME: (ptr noundef nonnull align 4 dereferenceable(4) [[THIS:%.*]]) unnamed_addr #[[ATTR1]] align 2 {
// CHECK1-NEXT:  entry:
// CHECK1-NEXT:    [[THIS_ADDR:%.*]] = alloca ptr, align 8
// CHECK1-NEXT:    store ptr [[THIS]], ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    [[THIS1:%.*]] = load ptr, ptr [[THIS_ADDR]], align 8
// CHECK1-NEXT:    ret void
//
//
// CHECK3-LABEL: define {{[^@]+}}@main
// CHECK3-SAME: () #[[ATTR0:[0-9]+]] {
// CHECK3-NEXT:  entry:
// CHECK3-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK3-NEXT:    [[REF_TMP:%.*]] = alloca [[CLASS_ANON:%.*]], align 1
// CHECK3-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK3-NEXT:    call void @"_ZZ4mainENK3$_0clEv"(ptr noundef nonnull align 1 dereferenceable(1) [[REF_TMP]])
// CHECK3-NEXT:    ret i32 0
//
//
// CHECK4-LABEL: define {{[^@]+}}@main
// CHECK4-SAME: () #[[ATTR1:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[RETVAL:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    store i32 0, ptr [[RETVAL]], align 4
// CHECK4-NEXT:    [[TMP0:%.*]] = load ptr, ptr getelementptr inbounds nuw ([[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], ptr @__block_literal_global, i32 0, i32 3), align 8
// CHECK4-NEXT:    call void [[TMP0]](ptr noundef @__block_literal_global)
// CHECK4-NEXT:    ret i32 0
//
//
// CHECK4-LABEL: define {{[^@]+}}@__main_block_invoke
// CHECK4-SAME: (ptr noundef [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca ptr, align 8
// CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca ptr, align 8
// CHECK4-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK4-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[BLOCK_ADDR]], align 8
// CHECK4-NEXT:    call void (ptr, i32, ptr, ...) @__kmpc_fork_call(ptr @[[GLOB4:[0-9]+]], i32 0, ptr @__main_block_invoke.omp_outlined)
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@__main_block_invoke.omp_outlined
// CHECK4-SAME: (ptr noalias noundef [[DOTGLOBAL_TID_:%.*]], ptr noalias noundef [[DOTBOUND_TID_:%.*]]) #[[ATTR3:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTGLOBAL_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK4-NEXT:    [[DOTBOUND_TID__ADDR:%.*]] = alloca ptr, align 8
// CHECK4-NEXT:    [[TMP:%.*]] = alloca ptr, align 8
// CHECK4-NEXT:    [[DOTOMP_IV:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[_TMP1:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[DOTOMP_LB:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[DOTOMP_UB:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[DOTOMP_STRIDE:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[DOTOMP_IS_LAST:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[G:%.*]] = alloca double, align 8
// CHECK4-NEXT:    [[G1:%.*]] = alloca double, align 8
// CHECK4-NEXT:    [[_TMP2:%.*]] = alloca ptr, align 8
// CHECK4-NEXT:    [[I:%.*]] = alloca i32, align 4
// CHECK4-NEXT:    [[BLOCK:%.*]] = alloca <{ ptr, i32, i32, ptr, ptr, double, ptr }>, align 8
// CHECK4-NEXT:    [[DOTOMP_REDUCTION_RED_LIST:%.*]] = alloca [2 x ptr], align 8
// CHECK4-NEXT:    store ptr [[DOTGLOBAL_TID_]], ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT:    store ptr [[DOTBOUND_TID_]], ptr [[DOTBOUND_TID__ADDR]], align 8
// CHECK4-NEXT:    [[TMP0:%.*]] = load ptr, ptr @g1, align 8
// CHECK4-NEXT:    store ptr [[TMP0]], ptr [[TMP]], align 8
// CHECK4-NEXT:    store i32 0, ptr [[DOTOMP_LB]], align 4
// CHECK4-NEXT:    store i32 1, ptr [[DOTOMP_UB]], align 4
// CHECK4-NEXT:    store i32 1, ptr [[DOTOMP_STRIDE]], align 4
// CHECK4-NEXT:    store i32 0, ptr [[DOTOMP_IS_LAST]], align 4
// CHECK4-NEXT:    store double 0.000000e+00, ptr [[G]], align 8
// CHECK4-NEXT:    [[TMP1:%.*]] = load ptr, ptr @g1, align 8
// CHECK4-NEXT:    store double 0.000000e+00, ptr [[G1]], align 8
// CHECK4-NEXT:    store ptr [[G1]], ptr [[_TMP2]], align 8
// CHECK4-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTGLOBAL_TID__ADDR]], align 8
// CHECK4-NEXT:    [[TMP3:%.*]] = load i32, ptr [[TMP2]], align 4
// CHECK4-NEXT:    call void @__kmpc_for_static_init_4(ptr @[[GLOB1:[0-9]+]], i32 [[TMP3]], i32 34, ptr [[DOTOMP_IS_LAST]], ptr [[DOTOMP_LB]], ptr [[DOTOMP_UB]], ptr [[DOTOMP_STRIDE]], i32 1, i32 1)
// CHECK4-NEXT:    [[TMP4:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK4-NEXT:    [[CMP:%.*]] = icmp sgt i32 [[TMP4]], 1
// CHECK4-NEXT:    br i1 [[CMP]], label [[COND_TRUE:%.*]], label [[COND_FALSE:%.*]]
// CHECK4:       cond.true:
// CHECK4-NEXT:    br label [[COND_END:%.*]]
// CHECK4:       cond.false:
// CHECK4-NEXT:    [[TMP5:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK4-NEXT:    br label [[COND_END]]
// CHECK4:       cond.end:
// CHECK4-NEXT:    [[COND:%.*]] = phi i32 [ 1, [[COND_TRUE]] ], [ [[TMP5]], [[COND_FALSE]] ]
// CHECK4-NEXT:    store i32 [[COND]], ptr [[DOTOMP_UB]], align 4
// CHECK4-NEXT:    [[TMP6:%.*]] = load i32, ptr [[DOTOMP_LB]], align 4
// CHECK4-NEXT:    store i32 [[TMP6]], ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND:%.*]]
// CHECK4:       omp.inner.for.cond:
// CHECK4-NEXT:    [[TMP7:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT:    [[TMP8:%.*]] = load i32, ptr [[DOTOMP_UB]], align 4
// CHECK4-NEXT:    [[CMP3:%.*]] = icmp sle i32 [[TMP7]], [[TMP8]]
// CHECK4-NEXT:    br i1 [[CMP3]], label [[OMP_INNER_FOR_BODY:%.*]], label [[OMP_INNER_FOR_END:%.*]]
// CHECK4:       omp.inner.for.body:
// CHECK4-NEXT:    [[TMP9:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT:    [[MUL:%.*]] = mul nsw i32 [[TMP9]], 1
// CHECK4-NEXT:    [[ADD:%.*]] = add nsw i32 0, [[MUL]]
// CHECK4-NEXT:    store i32 [[ADD]], ptr [[I]], align 4
// CHECK4-NEXT:    store double 1.000000e+00, ptr [[G]], align 8
// CHECK4-NEXT:    [[TMP10:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK4-NEXT:    store volatile double 1.000000e+00, ptr [[TMP10]], align 8
// CHECK4-NEXT:    [[BLOCK_ISA:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, double, ptr }>, ptr [[BLOCK]], i32 0, i32 0
// CHECK4-NEXT:    store ptr @_NSConcreteStackBlock, ptr [[BLOCK_ISA]], align 8
// CHECK4-NEXT:    [[BLOCK_FLAGS:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, double, ptr }>, ptr [[BLOCK]], i32 0, i32 1
// CHECK4-NEXT:    store i32 1073741824, ptr [[BLOCK_FLAGS]], align 8
// CHECK4-NEXT:    [[BLOCK_RESERVED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, double, ptr }>, ptr [[BLOCK]], i32 0, i32 2
// CHECK4-NEXT:    store i32 0, ptr [[BLOCK_RESERVED]], align 4
// CHECK4-NEXT:    [[BLOCK_INVOKE:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, double, ptr }>, ptr [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    store ptr @g1_block_invoke, ptr [[BLOCK_INVOKE]], align 8
// CHECK4-NEXT:    [[BLOCK_DESCRIPTOR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, double, ptr }>, ptr [[BLOCK]], i32 0, i32 4
// CHECK4-NEXT:    store ptr @__block_descriptor_tmp.1, ptr [[BLOCK_DESCRIPTOR]], align 8
// CHECK4-NEXT:    [[BLOCK_CAPTURED:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, double, ptr }>, ptr [[BLOCK]], i32 0, i32 5
// CHECK4-NEXT:    [[TMP11:%.*]] = load volatile double, ptr [[G]], align 8
// CHECK4-NEXT:    store volatile double [[TMP11]], ptr [[BLOCK_CAPTURED]], align 8
// CHECK4-NEXT:    [[BLOCK_CAPTURED4:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, double, ptr }>, ptr [[BLOCK]], i32 0, i32 6
// CHECK4-NEXT:    [[TMP12:%.*]] = load ptr, ptr [[_TMP2]], align 8
// CHECK4-NEXT:    store ptr [[TMP12]], ptr [[BLOCK_CAPTURED4]], align 8
// CHECK4-NEXT:    [[TMP13:%.*]] = getelementptr inbounds nuw [[STRUCT___BLOCK_LITERAL_GENERIC:%.*]], ptr [[BLOCK]], i32 0, i32 3
// CHECK4-NEXT:    [[TMP14:%.*]] = load ptr, ptr [[TMP13]], align 8
// CHECK4-NEXT:    call void [[TMP14]](ptr noundef [[BLOCK]])
// CHECK4-NEXT:    br label [[OMP_BODY_CONTINUE:%.*]]
// CHECK4:       omp.body.continue:
// CHECK4-NEXT:    br label [[OMP_INNER_FOR_INC:%.*]]
// CHECK4:       omp.inner.for.inc:
// CHECK4-NEXT:    [[TMP15:%.*]] = load i32, ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT:    [[ADD5:%.*]] = add nsw i32 [[TMP15]], 1
// CHECK4-NEXT:    store i32 [[ADD5]], ptr [[DOTOMP_IV]], align 4
// CHECK4-NEXT:    br label [[OMP_INNER_FOR_COND]]
// CHECK4:       omp.inner.for.end:
// CHECK4-NEXT:    br label [[OMP_LOOP_EXIT:%.*]]
// CHECK4:       omp.loop.exit:
// CHECK4-NEXT:    call void @__kmpc_for_static_fini(ptr @[[GLOB1]], i32 [[TMP3]])
// CHECK4-NEXT:    [[TMP16:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 0
// CHECK4-NEXT:    store ptr [[G]], ptr [[TMP16]], align 8
// CHECK4-NEXT:    [[TMP17:%.*]] = getelementptr inbounds [2 x ptr], ptr [[DOTOMP_REDUCTION_RED_LIST]], i64 0, i64 1
// CHECK4-NEXT:    store ptr [[G1]], ptr [[TMP17]], align 8
// CHECK4-NEXT:    [[TMP18:%.*]] = call i32 @__kmpc_reduce(ptr @[[GLOB2:[0-9]+]], i32 [[TMP3]], i32 2, i64 16, ptr [[DOTOMP_REDUCTION_RED_LIST]], ptr @__main_block_invoke.omp_outlined.omp.reduction.reduction_func, ptr @.gomp_critical_user_.reduction.var)
// CHECK4-NEXT:    switch i32 [[TMP18]], label [[DOTOMP_REDUCTION_DEFAULT:%.*]] [
// CHECK4-NEXT:      i32 1, label [[DOTOMP_REDUCTION_CASE1:%.*]]
// CHECK4-NEXT:      i32 2, label [[DOTOMP_REDUCTION_CASE2:%.*]]
// CHECK4-NEXT:    ]
// CHECK4:       .omp.reduction.case1:
// CHECK4-NEXT:    [[TMP19:%.*]] = load double, ptr @g, align 8
// CHECK4-NEXT:    [[TMP20:%.*]] = load double, ptr [[G]], align 8
// CHECK4-NEXT:    [[ADD6:%.*]] = fadd double [[TMP19]], [[TMP20]]
// CHECK4-NEXT:    store double [[ADD6]], ptr @g, align 8
// CHECK4-NEXT:    [[TMP21:%.*]] = load double, ptr [[TMP1]], align 8
// CHECK4-NEXT:    [[TMP22:%.*]] = load double, ptr [[G1]], align 8
// CHECK4-NEXT:    [[ADD7:%.*]] = fadd double [[TMP21]], [[TMP22]]
// CHECK4-NEXT:    store double [[ADD7]], ptr [[TMP1]], align 8
// CHECK4-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP3]], ptr @.gomp_critical_user_.reduction.var)
// CHECK4-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK4:       .omp.reduction.case2:
// CHECK4-NEXT:    [[TMP23:%.*]] = load double, ptr [[G]], align 8
// CHECK4-NEXT:    [[TMP24:%.*]] = atomicrmw fadd ptr @g, double [[TMP23]] monotonic, align 8
// CHECK4-NEXT:    [[TMP25:%.*]] = load double, ptr [[G1]], align 8
// CHECK4-NEXT:    [[TMP26:%.*]] = atomicrmw fadd ptr [[TMP1]], double [[TMP25]] monotonic, align 8
// CHECK4-NEXT:    call void @__kmpc_end_reduce(ptr @[[GLOB2]], i32 [[TMP3]], ptr @.gomp_critical_user_.reduction.var)
// CHECK4-NEXT:    br label [[DOTOMP_REDUCTION_DEFAULT]]
// CHECK4:       .omp.reduction.default:
// CHECK4-NEXT:    call void @__kmpc_barrier(ptr @[[GLOB3:[0-9]+]], i32 [[TMP3]])
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@g1_block_invoke
// CHECK4-SAME: (ptr noundef [[DOTBLOCK_DESCRIPTOR:%.*]]) #[[ATTR2]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTBLOCK_DESCRIPTOR_ADDR:%.*]] = alloca ptr, align 8
// CHECK4-NEXT:    [[BLOCK_ADDR:%.*]] = alloca ptr, align 8
// CHECK4-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[DOTBLOCK_DESCRIPTOR_ADDR]], align 8
// CHECK4-NEXT:    store ptr [[DOTBLOCK_DESCRIPTOR]], ptr [[BLOCK_ADDR]], align 8
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, double, ptr }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 5
// CHECK4-NEXT:    store double 2.000000e+00, ptr [[BLOCK_CAPTURE_ADDR]], align 8
// CHECK4-NEXT:    [[BLOCK_CAPTURE_ADDR1:%.*]] = getelementptr inbounds nuw <{ ptr, i32, i32, ptr, ptr, double, ptr }>, ptr [[DOTBLOCK_DESCRIPTOR]], i32 0, i32 6
// CHECK4-NEXT:    [[TMP0:%.*]] = load ptr, ptr [[BLOCK_CAPTURE_ADDR1]], align 8
// CHECK4-NEXT:    store double 2.000000e+00, ptr [[TMP0]], align 8
// CHECK4-NEXT:    ret void
//
//
// CHECK4-LABEL: define {{[^@]+}}@__main_block_invoke.omp_outlined.omp.reduction.reduction_func
// CHECK4-SAME: (ptr noundef [[TMP0:%.*]], ptr noundef [[TMP1:%.*]]) #[[ATTR5:[0-9]+]] {
// CHECK4-NEXT:  entry:
// CHECK4-NEXT:    [[DOTADDR:%.*]] = alloca ptr, align 8
// CHECK4-NEXT:    [[DOTADDR1:%.*]] = alloca ptr, align 8
// CHECK4-NEXT:    store ptr [[TMP0]], ptr [[DOTADDR]], align 8
// CHECK4-NEXT:    store ptr [[TMP1]], ptr [[DOTADDR1]], align 8
// CHECK4-NEXT:    [[TMP2:%.*]] = load ptr, ptr [[DOTADDR]], align 8
// CHECK4-NEXT:    [[TMP3:%.*]] = load ptr, ptr [[DOTADDR1]], align 8
// CHECK4-NEXT:    [[TMP4:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 0
// CHECK4-NEXT:    [[TMP5:%.*]] = load ptr, ptr [[TMP4]], align 8
// CHECK4-NEXT:    [[TMP6:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 0
// CHECK4-NEXT:    [[TMP7:%.*]] = load ptr, ptr [[TMP6]], align 8
// CHECK4-NEXT:    [[TMP8:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP3]], i64 0, i64 1
// CHECK4-NEXT:    [[TMP9:%.*]] = load ptr, ptr [[TMP8]], align 8
// CHECK4-NEXT:    [[TMP10:%.*]] = getelementptr inbounds [2 x ptr], ptr [[TMP2]], i64 0, i64 1
// CHECK4-NEXT:    [[TMP11:%.*]] = load ptr, ptr [[TMP10]], align 8
// CHECK4-NEXT:    [[TMP12:%.*]] = load double, ptr [[TMP7]], align 8
// CHECK4-NEXT:    [[TMP13:%.*]] = load double, ptr [[TMP5]], align 8
// CHECK4-NEXT:    [[ADD:%.*]] = fadd double [[TMP12]], [[TMP13]]
// CHECK4-NEXT:    store double [[ADD]], ptr [[TMP7]], align 8
// CHECK4-NEXT:    [[TMP14:%.*]] = load double, ptr [[TMP11]], align 8
// CHECK4-NEXT:    [[TMP15:%.*]] = load double, ptr [[TMP9]], align 8
// CHECK4-NEXT:    [[ADD2:%.*]] = fadd double [[TMP14]], [[TMP15]]
// CHECK4-NEXT:    store double [[ADD2]], ptr [[TMP11]], align 8
// CHECK4-NEXT:    ret void
//